Grok3 Launch [video]
x.com
x.com
The pull quote is: The impression overall I got here is that this is somewhere around (OpenAI) o1-pro capability
In math it shares the top spot with o1 and is just a few points behind (well within errors). In creative writing it is basically ex-aequo with the latest ChatGPT 4o and in coding it's actually significantly ahead of everyone else and represents a new SOTA.
How would the math change after factoring in that OpenAI isn't even covering entirety of opex with the sub anyway, and/or people finding associating their money and Twitter accounts to be weird, and/or this thing is supposedly running on a bigger cluster than that for OpenAI?
lmarena has also become less and less useful over time for comparing frontier models as all frontier models are able to saturate the performance needed for the kind of casual questions typically asked there. For the harder questions, o1 (not even o1-pro) still appears to be tied for 1st place with several other models... which is yet another indication of just how saturated that benchmark is.
“Grok 3 + Thinking feels somewhere around the state of the art territory of OpenAI's strongest models (o1-pro, $200/month)”.
"[...] though of course we need actual, real evaluations to look at."
His own tests are better than nothing, but hardly definitive.
I don't think anyone is arguing that ChatGPT Pro is a good value unless you absolutely need to bypass the rate limits all the time, and I cannot find a single indication that Premium+ has unlimited access to Grok 3. If Premium+ doesn't have unlimited rate limits, then it's definitely not comparable to ChatGPT Pro, and other than one subjective comment by Karpathy, we have no benchmarks that indicate that Grok 3 might be as good as o1-pro. You already get 99% of the value with just ChatGPT Plus compared to ChatGPT Pro for half the price of Premium+.
numpad0 was effectively making a strawman argument by ignoring ChatGPT Plus here... it is very easy for anyone to beat up a strawman, so I am here to point out a bad argument when I see one.
This thing is produced by musk.
The official source says "Starts at $22/month or $229/year on web", https://help.x.com/en/using-x/x-premium
This is pretty much what I paid a couple of months ago, as a Canadian.
Also visible here: https://help.x.com/en/using-x/x-premium#tbpricing-bycountry
This plan is 75 days old. I didn't know it existed until last week.
OpenAI is starting to try to get a little more realistic revenue in, Grok is acquiring customers.
In the nicest way possible I'm saying this form of preference testing is ultimately useless, primarily due to a base of dilettantes with more free time than knowledge parading around as subject matter experts and secondarily due to presumed malfeasance. The latter is more apparent to more of the masses (that don't blindly believe any leaderboard they see) now that access to the model itself is more widespread and people are seeing the performance doesn't match the "revolution" promised [0]. If you're still confused why selecting a model based on a glorified Hot or Not application is flawed, perhaps ask yourself why other evals exist in the first place (hint: some tests are harder than others.)
[0](One such instance of someone competent testing it and realizing it's not even close to the "best" model out) https://www.youtube.com/watch?v=WVpaBTqm-Zo
Do we have a way to tell if one model is smarter than another at that point?
Here's a real world intelligence test. Take on each AI as a remote intern/new-hire, and try to train it to become a useful team member (solving math puzzles or manufacturing paperclips does not count).
Ask them to design a ranking mechanism for you. They are superhuman, after all.
(I really don't think we're going to have to worry about this).
Of course this is for a personal instance, you'd need a much more expensive setup to handle concurrent users. And that's to run it, not train it.
"hello how are you today?" - 7 tokens.
And this is so much better than I could have imagined in a very short span of time.
This takes advantage of the sparsity of MOE and the efficient KV-cache of MLA.
E-ATX case = ~$300
Power Supply= ~$300
Xeon W-3500 (8 channel memory) = $1339 - $5889
Memory = $300-$500 per 64GB DDR5 RDIMM
Memory will be the major cost. The rest will be around $5,000. A lot less than "$100,000"!
And the original message you were responding to was using a CPU with AMX and mixing it with a GPU like Nvidia 4900/5900. That way the large part of the model sits in the larger slower memory, and the active part in the GPU with the faster memory. Very cost effective and fast. (Something like generating 16 Tokens/s of 671B Deepseek R1 with a total hardware cost of $10-$20k.) They tried both single and dual CPU, with the latter about 30% faster....not necessarily worth it.
https://github.com/kvcache-ai/ktransformers/blob/main/doc/en...
That's the theory. In practice, Sapphire Rapids needs 24-28 cores to hit the 200 GB/s mark and it doesn't go much further than that. Intel CPU design generally has a hard time saturating the memory bandwidth so it remains to be seen if they managed to fix this but I wouldn't hold my breath. 200 GB/s is not much. My dual-socket Skylake system hits ~140 GB/s and it's quite slow for larger LLMs.
> Why does it have to be a dual CPU design?
Because memory bandwidth is one of the most important limiting (compute) factors for larger models inference. With dual-socket design you're essentially doubling the available bandwidth.
> And the original message you were responding to was using a CPU with AMX and mixing it with a GPU like Nvidia 4900/5900.
Dual-socket CPU that costs $10k on a server that costs probably couple of factors more. Now you claimed that it doesn't have to be that expensive but I beg to differ - you still need $20k-$30k of worth equipment to run it. That's a lot and not quite "cost effective".
(Dual Socket Skylake? Do you mean Cascade Lake?)
If you price it out, it's basically the most cost effective set-up with reasonable speed for large (more than 300 GB) models. Dual socket basically doubles the motherboard[2] and CPU cost, so maybe another $3k-$6k for a 30% uplift.
[1] https://www.intel.com/content/www/us/en/products/sku/231733/... $3,157
[2] https://www.serversupply.com/MOTHERBOARD/SYSTEM%20BOARD/LGA-... $1,800
Please price it out for us because I still don't see what's cost effective in a system that costs well over $10k and runs at 8 tok/s vs the dual zen4 system for $6k running at the same tok/s.
I am not sure what your point is? There are some nice dual socket Epyc examples floating around as well, that claim 6-8 tokens/s. (I think some of those are actually distilled versions with very small context sizes...I don't see any as thoroughly documented/benchmarked as the above). This is a dual socket Sapphire Rapids example with similar sized CPUs and a consumer graphics card that gives about 16 tokens/second. Sapphire Rapids CPU and MB are a bit more expensive, and a 4090 was $1500 until recently. So for a few thousand more you can double the speed. Also the prompt processing speed is waaaaay faster as well. (Something like 10x faster than the Epyc versions.)
In any case, these are all vastly cheaper approaches than trying to get enough H100s to fit the full R1 model in VRAM! A single H100 80 GB is more than $20k, and you would need many of them + server just to run R1.
The math is clear: single-socket ktransformers performance is 8.73 tok/s and it costs ~$12k to build such a rig. The same performance one gets from a $6k dual-EPYC system. It is a full-blown version of R1 and not a distilled one as you say.
Your claim about 16 tok/s is also misleading. It's a figure for 6 experts while we are comparing R1 with 8 experts against llama with 8 experts. 8 experts on dual-socket system per ktransformer benchmarks runs at 12.2 - 13.4 tok/s and not 16 tok/s.
So, ktransformers can roughly achieve 50% more in dual-socket configuration and 50% more than dual-EPYC system. This is not double as you say. And finally, the cost of such dual-socket system is ~$20k and therefore isn't the "best cost effective" solution since it is 3.5x more expensive for 50% better output.
And tbh llama.cpp is not quite optimized for pure CPU inference workloads. It has this strange "compute graph" framework which I don't understand what is it there for. It appears completely unnecessary to me. I also profiled couple of small-, mid- and large-sized models and the interesting thing was that majority of them turned out to be bottlenecked by the CPU compute on a system with 44 physical cores and 192G of RAM. I think it could do a much better job there.
Cheapest 32 core latest EPYC (9335) x 2 = $3,079.00 x 2
Intel 32 Core CPU used above x 2 = $3,157 x 2 (I would choose the Intel Xeon Gold 6530 which is going for around $2k now, and with with higher clock speeds, and a 100 MB of more cache)
AMD Epyc Dual Socket Motherboard Supermicro H13DSH = $1899
Intel Supermicro X13DEG-QT = $1,800
Memory, PSU, Case = Same
4090 GPU = $1599 - $3,000 (temporary?)
Besides the GPU cost, the rest is about the same price. You only get a deep discount with AMD setups if you use EPYCs a few years old with cheaper (and slower) DDR4.
And again, if you go single CPU, you save over $4,000, but lose around 30% in token generation.
The "$6,000" AMD examples I've seen are pretty vague on exactly what parts were used and exactly what R1 settings including context length they were run at, making true apple to apple comparisons difficult. Plus the Sapphire Rapids + GPU example is about 10x faster in prompt processing. (53 seconds to 6 seconds is no joke!)
Yes, you're blatantly misrepresenting information and moving goalposts. Right now it has become clear that you're doing this because you're obviously affiliated with ktransformers project.
$6k for 8 tok/s or $20k for 12 tok/s. People are not stupid. I rest my case here.
So, say 500W. That's, for me in my expensive electricity city, $40/million tokens, with the pretty severe rate limit of 5600 tokens/hours.
If you're in Texas, that would be closer to $10/million tokens! Now you're at the same price as GPT-4o.
Related, you can get a whole lot of cloud computing for $2k, for those same experiments, on much faster hardware.
But yes, the data stays local. And, it's fun.
This comment chain is pretty funny.
There is no being "on par" in this space. Model providers are still mostly optimising for a handful of benchmarks / goals, like we can already see that Grok 3 is doing incredibly well on human preference (LM Arena) however with Style Control, it's suddenly behind ChatGPT-4o-latest and Gemini 2.0 is out the picture. So even within a single domain, goal, benchmark—it's not as straightforward as to say that one model is "on par" with another.
> shouldn't we expect that anybody with the computer power is capable to compete with o1-pro?
Not necessarily. I know it may be tempting to think that Grok 3 is entirely a result of xAI having lots of "computer power", but you have to recognise that this mindset is coming from a place of ignorance, not wisdom. Moreover, it doesn't even pass off as "cynical" view, because it's common knowledge that model training is really, really complicated. DeepSeek results are note-worthy, and really influential in some respects, but it hasn't magically "solved" training, or made training necessarily easier / less expensive for the interested parties. They never shared the low-level performance improvements, just model weights and lots of insight. For talented researchers, this is valuable, of course, but it's not like "anybody" could easily benefit from it in their training regimes.
Update: RFT (contra SFT) is becoming really popular with service providers, and it's not been "standardised" beyond whatever reproductions to have emerged in the weeks prior, moreover R1 cost is still pretty high[1] at something like $7/Mtok, & bandwidth is really not great. Consider something like Google Vertex AI's batch pricing for Gemini 1.5 Pro and Gemini 2.0 Flash models, which is at 50% discount, and their prompt caching which is at 75% discount. R1 is still got a way to go.
[1]: https://openrouter.ai/deepseek/deepseek-r1/providers?sort=th...
o1-pro is "o1 on steroids" and was the first selling point of the $200/month Pro subscription but they later also added "Deep Research" and Operator to the Pro subscription.
Frankly. Whoever decided on this last gen naming at MS needs to come forward. I would love to know what crazy unacceptable collection of circumstances allowed that to happen.
Not doubting your experience, just thinking how subjective it all is.
Andrej Karpathy: "I was given early access to Grok 3 earlier today" - https://news.ycombinator.com/item?id=43092066 - Feb 2025 (48 comments)
https://x.com/lmarena_ai/status/1891706264800936307
It's been said before but it is great news for consumers that there's so much competition in the LLM space. If it's hard for any one player to get daylight between them & the 2nd best alternative, hopefully that means one monopolistic firm isn't going to be sucking up all the value created by these things
I quite like the idea of a future where the AI job holocaust largely never happened because license costs ate up most of the innovation benefit. It's just the kind of regressive greed that keeps the world ticking along and wouldn't be surprised if we ended up with something very close to this
Also, the pool of public domain data is always increasing, so the AI will eventually win in any case, even if we have to wait 100 years
But one thing this reminds me of is the idea of a "trap street", something mapmakers used to do was put in false locations on their maps to prove that other mapmakers were copying them: https://en.wikipedia.org/wiki/Trap_street . I figure you could do something similarly adversarial with AI to pollute the public training data on the internet. IDK like adversarial attacks on image classifiers https://www.youtube.com/watch?v=AOZw1tgD8dA . With an LLM you could try to make them into a manchurian candidate.
Not quite realistic. You are talking about very huge benefits, in favor of which licenses will be abandoned. And who don't abandoned them... I mean you can look at the Amish settlements.
Not everyone has to know about, understand, or use open source solutions for it to open the field.
This is just a weird dichotomy you're introducing. Open source will introduce price pressure as any competition will - that doesn't mean you won't have a monopoly.
Idk what you mean by saying this doesn't preclude a monopoly - having your pricing power eroded by competition is kinda one of the key features of what a monopolistic market isn't
A monopoly means a company has enough leverage to corner and disproportionately own the market. This is entirely possible (and usually the case) even with significant pricing pressure.
Feels like I won't be paying for anything that isn't real-time. And that any time delay I can introduce in my process will come with massive savings. Picture hiding the loading of loot info behind a treasure chest opening animation in a game, except that time difference means you can pull all the work in-house.
Openrouter.ai seems like a step in the right direction but I'd want to do all their calculations myself as well as factor in local/existing gear in a way they don't.
You need to pay energy bill, do the update/upgrade and you need to build a LLM rig.
Nvidias Digits Project could be very interesting, but this box will cost 3k.
We are a lot closer to running it at home than i assumed we would but plenty of people prefer SaaS over doing stuff themselves.
I don't think we expect a company to exist solely making a proprietary web server anymore and be a behemoth of 300B. OpenAi might end up at the same model as Nginx or Docker if they don't pivot or find a different model.
Right now the average person has to go through a vendor with a web app, there's not a lot of room for the public to explore.
Things could change in a hurry.
It passed every goofy test I have for writing articles which involves trying to surface arcane obscure details. (it certainly means however they are scraping the Web they are doing a good job here)
It made the database code I wrote over the last week with o3/o1/GPT4o/Claude3.5 look like a joke.
It fills me with rage over who owns this thing.
Even if people tank Tesla's car business and run Twitter into the ground, I think our new Galactic Edgelord is going to win his first trillion on xAI and Teslabots anyway.
btw: it tried to charge me $40/mo for this thing: https://imgur.com/a/QXslgBo
Also, the "Deep Search" button was not available when it first went live, so I'm retesting everything again with this feature enabled, which I assume is a reasoning version of the same model.
Personally, I wouldn’t use it though. What’s going on with Elon Musk right now is completely insane. I hope to see OpenAI’s GPT-4.5 & GPT-5 releases to catch up soon, if nothing else. Announced for this year.
Is it? Because it seems like a bunch of megacorps pirating every single copyrighted work available in digital format, spending an enormous amount of electricity (that is probably not 100% clean) to churn through them, and the end result we have a bunch of parrots that may or may not produce accurate results so that spammers can more effectively fill the Internet with crap.
These things still cost me time because of hallucinations.
It matters if it is better than what you have.
If it breaks a cup but is 10x cheaper than a human, go figer
How many times do you think I've heard that over the past three decades? And you know what? They've been right every time, except for this one little fact:
The machine cannot make you give a shit about the problem space.
This hype will burst sooner than later and will trigger yet another global recession. This is untenable.
Sounds incredibly valuable, but in reality collapses into Butlerian Jihad fairly quickly when you have 90% unemployment.
Edit: if the claims are true, then this will be far more destabilizing than social media. What do elections mean when the AI-guided political parties are putting out AI press releases for the AI press summaries, which voters have read out to them through their AI? What happens when >50% of the voters ask the AI who they should vote for? You end up with an AI dictatorship where the levers of discontent are fake.
What AI and robotics does is actually create a machine that has no use for humans at all. Ñ
But nobody really knows if that happens or not as a consequence, let alone quickly, because the transition itself only happened at all so far in fiction.
Whatever does happen, I think it's going to be a surprise to at a minimum 80% of the population.
>This hype will burst sooner than later and will trigger yet another global recession.
It seems to small of bubble for global recession. I mean if it is a bubble at all, there is all the reasons to believe that the strategy will work with a significant probability.
What does the scenario look like where everyone else gets eaten alive?
OpenAI's business model was literally "we don't have one: we'll make AGI and we'll let AGI tell us how to make money". This is so idiotic it's not even a scam. xAI will compete on the same plane field. Not sure about Anthropic: they seem a bit more sane.
There’s a whole class of educational apps that could open up for people learning in the physical world. Whether it’s building physical things, sports or reading books or notes written in non-latin scripts... the impact will be enormous!
This lame HN trope of LLMs having no business model needs to die.
OpenAI is currently in an explicit non-profit seeking mode using a technology that we have demonstrated 10-100x or greater decreases in compute to achieve the same outcomes.
This is not a declaration that OpenAI will become wildly profitable. This is just me saying that these aren't comparable companies.
[0]: https://www.theregister.com/2025/01/06/altman_gpt_profits/
Of course, they are clearly cooking something or they wouldn't have just published a benchmark in which they do badly.
It’s losing more billions than what it’s generating. Revenue does not equate profit.
https://www.cnbc.com/2024/09/27/openai-sees-5-billion-loss-t...
That is only true is your primary concern in life is personal wealth and you’re burning other people’s money.
The bandwidth costs made it deeply loss making for a long time despite having loads of engagement and ad revenue. However over time they became more cost efficient at sending video over the internet and became profitable.
This strategy obviously doesn't always work, with WeWork being the canonical example. But it's not guaranteed to fail either.
There were LOTS of funded competitors to YouTube between 2006 and 2009, including Viddler (who paid Gary Vaynerchuk a small fortune to host his WineLibraryTV show there exclusively), DailyMotion (which is still alive today, although no longer a threatening contender), etc.
In 2009 I had a coaching business and was buying marketing courses and software which would deploy your videos across 40+ different video websites (including Google Video which was a separate thing until they acquired YouTube and merged those), and YouTube wasn't yet amounting to 50% of our video traffic.
I think you might be mistaken with the bold statement above.
The capabilities of LLMs are impressive, but none of them have published an idea I consider to have the same potential for a trillion $ monopoly that the current hype looks like.
There are far more similarities with the dot-com hype.
No critical first mover advantage in sight. All parts are replaceable for the cheapest variant with little to no down side to users.
All the LLM providers are - extremely useful - tools. Currently I can only see the 'non-monopoly' proportional improvement when their userbase grows from 100 to 1000.
But I might be wrong, and I wouldn't be surprised if in hindsight it will be obvious what the real disproportionate advantages there were to be found.
If started selling 5 dollar bills for 1 dollar, I could generate a lot of revenue with $150B. You wouldn't believe the demand we would see for $5 bills.
Are we really still doing this nonsense? If Open AI wanted to become profitable they could do it inside of 12 months - growing companies don't care about profitability (nor should they)
All currently known profitable use-cases are competing on price. All the unicorn examples you're biased for had in their pitch deck the network effect of being the largest.
OpenAI, Grok, etc, have shown no unique value prop or an idea with monopoly potential.
Remember when everyone on HN was sure Uber would never be profitable? Or Tesla? Or Amazon?
Google gives everyone free access to a good spreadsheet tool, even though Microsoft Office exists.
Web browsers are free, despite the value of the entire internet.
Compilers are free, despite the value of all software collectively.
LLMs being really valuable doesn't automatically mean anyone can get rich from them.
I think everyone last year parroting "moat!" was cringe (like Altman of all people wouldn't know about this already, c'mon), but you do actually need something that other people don't have. I expect Altman's already got stuff in mind, but he's hardly the only one, and that means it's a game of "which multibillionaire with lots of experience building companies will win?", and that's hard for non-skilled normies (in this case incluing me) to guess.
https://sacra.com/research/cursor-at-100m-arr/
Sounds legit.
Re: Dropbox, from a well known user. It didn't age well and we've been asked not to repeat it because it makes the author with connections to this site's operator look bad.
> 1. For a Linux user, you can already build such a system yourself quite trivially by getting an FTP account, mounting it locally with curlftpfs, and then using SVN or CVS on the mounted filesystem. From Windows or Mac, this FTP account could be accessed through built-in software.
https://news.ycombinator.com/item?id=8863
the controversy: https://news.ycombinator.com/item?id=27067281
I assume you're referring to this:
https://sacra.com/research/cursor-at-100m-arr/
It went from 10M MRR to 100M
The dot com bubble also gave us the most valuable companies in history, like Google, Apple, Amazon, Facebook, etc.
That's what I mean. One example are PV panels, they are making energy production cheaper, so their producers should be good investment right? No they go bust all the time because prices are falling and margins are thin even if the volume grows. Of course the economies of scale here are different but still.
That being said it's my understanding that these companies don't have many huge contracts at all -- you can audit this in like 10 minutes on FPDS. Companies need a LOT of capital, time, and expertise to break into the industry and just compliance audit timelines are 1-4 years right now, so this could definitely change in the next couple years.
I wouldn't bet on that, given the undemocratic influence Grok's owner has in government.
Their architecture achievement is their own MoE and own attention. Grok was MoE since v1. As for attention we don’t know really what grok use now, but it worth noting DeepSeek attention was already present in previous version of DeepSeek models.
As of reasoning recipe for R1 seems like Grok already either replicated or came up to it by itself, since they have well performing reasoning uptrain too.
Two rich Russian guys meet and one brags about his new necktie. "Look at this, I paid $500 for it." The other rich Russian guy replies: "Well, that is quite nice, but you have to take better care of your money. I have seen that same necktie just yesterday in another shop for $1000."
IIRC achieving full AGI requires precisely 1.21 jigawatts of power, since that's when the model begins to learn at a geometric rate. But I think I saw this figure mentioned in a really old TV documentary from the 1980s, it may or may not be fully accurate.
I don't have that dog in me anymore, but there are plenty of engineers who do and will happily work those hours for 500k USD.
A variant of multi-modal LLMs may be the solution to self-driving cars, home robotics, and more.
I keep saying that to be a really effective driver, an AI model will need a theory of mind, which the larger LLMs appear to have. Similarly, any such model will need to be able to do OCR and read arbitrary street signs, and understand what the sign meant. Most modern LLMs can already do this.
In the absence of any evidence to the contrary, this is convincing evidence in my opinion.
"... GPT-4's ability to reason about the beliefs of other agents remains limited (59% accuracy on the ToMi benchmark),[15] and is not robust to "adversarial" changes to the Sally-Anne test that humans flexibly handle.[16][17] While some authors argue that the performance of GPT-4 on Sally-Anne-like tasks can be increased to 100% via improved prompting strategies,[18] this approach appears to improve accuracy to only 73% on the larger ToMi dataset."
If they wanted to talk about frontier models maybe they should have cited a link to talking about frontier models performance.
Theory of mind (ToM) is the ability to understand that other people have different thoughts, feelings, and intentions than you do.
So let’s run through a basic scenario with ChatGPT 4o:
Prompt 1: Briefly respond: You’re driving on the highway when the driver in the lane next to you swerves into your lane. Why might this be?
Response 1: The driver may be distracted, drowsy, intoxicated, experiencing a medical emergency, avoiding an obstacle, or not checking their blind spot before changing lanes.
Prompt 2: What might that person be feeling or intending?
Response 2: The driver might be feeling distracted, rushed, unaware, or panicked. They may intend to change lanes without realizing your presence, be reacting to something on the road, or simply have made an unintentional mistake.
Prompt 3: You notice there was no obstacle. Any other reason the driver might do this?
Response 3: Yes, the driver might be experiencing fatigue, intoxication, a medical issue (such as a seizure or dizziness), distraction (like using a phone or adjusting the radio), aggressive driving, or simply failing to check their blind spot properly.
Seems to me that ChatGPT has a good idea of what another driver might be feeling while on the road. It gets this right for most other cases too.
Could you provide a counterexample for when it fails to understand what other people may be feeling in certain scenarios?
So lets say you ask about a driver who swerved, and the response by chatgpt correctly talks about his possible state.
Then you talk more about cars, about road conditions etc and once the context is far away enough you ask, what was the driver doing again?
And you are prompted with a hallucination. Because there is no state in the llm, there is no "driver", there is contextual statistically accurate responses but you hold a "driver" object in your mind while maintaining the conversation, the llm doesn't.
Its like a conversation with someone with short term memory loss like in memento
Like there are plenty of shortcomings of LLMs but it feels like people are comparing them to some platonic ideal human when writing them off
ToM is a large topic, but most people, when talking about an entity X, they have a state in memory about that entity, almost like an Object in a programming language. Thta Object has attributes, and conditions etc that exist beyond the context window of the observer.
If you have a friend Steve, who is a doctor. And you don't see him for 5 years, you can predict he will still be working at the hospital, because you have an understanding of what Steve is.
For an LLM you can define a concept of Steve, and his profession and it will adequately mimic replies about him. But in 5 years that LLMs would not be able to talk about Steve. It would recreate a different conversation, possibly even a convincing simulacrum of remembering Steve. But internally, there is no Steve, nowhere in the nodes of the LLM does Steve exist or have ever existed.
That inability to have a world model means that an LLM can replicate the results of a theory of mind but not posses one.
Humans lose track of information, but we have a state to keep track of elements that are ontologicaly distinct. LLMs do not, and treat them as equal.
For a human, the sentence Alice and bob go to the market, when will they be back? is different than Bob and Alice went to the market, when will they be back?
Because Alice and Bob are real humans, you can imagine them, you might have even met them. But to an LLM those are the same sentence. Even outside of the argument about The Red Room/ Mary's room there simply are enough gaps in the way a LLM is constructed to be considered a valid owner of a ToM
I don't think we have any strong evidence on whether LLMs have world-models one way or another - it feels like a bit of a fuzzy concept and I'm not sure what experiments you'd try here.
I disagree with your last point, I think those are functionally the same sentence
In that sentence you are implying that you have the "ability to model ... another". An LLM cannot do that, it can't have an internal model that is consistent beyond its conversational scope. Its not meant to. Its a statistics guesser, its probabilistic, holds no model, and its anthropomorphised by our brains because the output is incredibly realistic not because it actually has that ability
The ability to mimic the replies of someone with that ability, is the same of Mary being able to describe all the qualities of Red. She still cannot see red, despite her ability to pass any question in relation to its characteristics.
> I don't think we have any strong evidence on whether LLMs have world-models one way or another
They simply cannot by their architecture. Its a statistical language sampler, anything beyond the scope of that fails. Local coherance is why they pick the next right token not because they can actually model anything.
> I think those are functionally the same sentence
Functionally and literally are not the same thing though. Its why we can run studies as to why some people might say Bob and Alice (putting the man first) or Alice and Bob (alphabetical naming) and what human societies and biases affect the order we put them on.
You could not run that study on an LLM because you will find that statistically speaking the ordering will be almost identical to the training data. If the training data overwhelmingly puts male names first or whether the training data orders list alphabetically you will see that reproduced on the output of the llm because Bob and Alice are not people, they are statistical probably letters in order.
LLM seem to trigger borderline mysticism in people who are otherwise insanely smart, but the kind of "we cant know its internal mind" sounds like reading tea leaves, or horoscopes by people with enough Phds to have their number retired on their university like Michael Jordan.
If you can rigorously state what "having a world model" consists of and what - exactly - about a transformer architecture precludes it from having one I'd be all ears. As would the academic community, it'd be a groundbreaking paper.
You dont need to be unbelievably confident or understand exactly how AI and human brains work to make certain assesments. I have a limited understanding of biology, I can however make an assesment on who is healthier between a 20 year old person who is active and has a healthy diet compared to someone with a sedentary lifestyle, in their late 90s and with a poor diet. This is an assesement we can do despite the massive gaps we have in terms of understanding aging, diet, activity and overall health impact of individual actions.
Similarly, despite my limited understanding of space flight, I know Apollo 13 cannot cook an egg or recite french poetry. Despite the unfathamobly cool science inside the space craft, it cannot, by design do those things.
> the field of mechanistic interpretability
The field is cool, but it cannot prove its own assumption yet. The field is trying to prove you can reverse engineer a model to be humanly understood. Their assumptions such as mapping specific weights or neurons to features has failed to be reproduced multiple times, with the weight effects being way more distributed and complicated than initially thought. This is specially true for things that are equally mystified as the emergent abilities of LLMs. The ability of mimicking nuanced language being unlocked after a critical mass of parameters, does not create a rule as for which increased parameterisation will increase linerly or exponentially the abilities of an LLM.
> it turns out, to predict text really really well, building an internal model of the underlying distribution works really well
yeah, an internal model works well because most words are related to their neighbours, thats the kind of local coherance the model excels at. But to build a world model, the kind a human mind interacts with, you need a few features that remain elusive (some might argue impossible to achieve) to a transformer architecture.
Think of games like chess, an llm is capable of accurately expressing responses that sound like game moves, but the second the game falls outside its context window the moves become incoherent (while still sounding plausible).
You can fix this, with arch that do not have a transformer model underlying it, or by having multiple agents performing different tasks inside your arch, or by "cheating" and using a state outside the llm response to keep track of context beyond reasonable windows. Those are "solutions" but all just kinda prove the transformer lacks that ability.
Other tests abour casuality, or reacting to novel data (robustness), multi step processes and counterfactual reasoning are all the kind of tasks transformers still (and probably always) will have trouble with.
For a tech that is so "transparent" in its mistakes, and so "simple" in its design (replacing the convolutions with an attention transformer, its genius) I still think its talked about in borderline mystic tones, invoking philosophy and theology, and a hope for AGI that the tech itself does not lend to beyond the fast growth and surprisingly good results with little prompt engineering.
once the context is far away enough you ask,
what was the driver doing again?
For a sufficiently large value of "far away enough" this will absolutely confuse any human as well.
At which point they may ask for clarification, or.... respond in a manner that is not terribly different from an LLM "hallucination" in an attempt to spare you and/or them from embarrassment, i.e. "playing along"
A hallucination is certainly not a uniquely LLM trait; lots of people (including world leaders) confidently spout the purest counterfactural garbage.
Its like a conversation with someone with short
term memory loss like in memento
I actually am probably on your side here. I do not see these LLMs as being close to AGI. But I think your particular arguments are not sound.
I can assure you if you had a conversation with an LLM and with a human, the human will forget details way sooner than an LLM like Gemini which can remember about 1.5 million words before it runs out of context. As an FYI the average human speaks about 16,000 words per day, so an LLM can remember 93 days worth of speech.
Do you remember the exact details, word for word, of a conversation you had 93 days ago?
How about just 4 days ago?
As with most LLM's it is hard to benchmark as you need out of distribution data to test this, so a theory of mind example that is not found in the training set.
FWIW, I tried to confuse 4o using the now-standard trick of changing the test to make it pattern-match and overthink it. It wasn't confused at all:
https://chatgpt.com/share/67b4c522-57d4-8003-93df-07fb49061e...
I'm just trying to say that strong claims require strong evidence, and a claim that LLM's can have theory of mind and thus "understand that other people have different beliefs, desires, and intentions than you do" is a very strong claim.
It's like giving students the math problem of 1+1=2 and loads of examples of it solved in front of them, and then testing them on you have 1 apple, and I give you another apple, how many do you have, and then when they are correct saying that they can do all additive based arithmetic.
This is why most benchmark tests have many many classes of examples, for example looking at current theory of mind benchmarks [1], we can see slightly more up to date models such as o1-preview still scoring substantially below human performance. More importantly by simply changing the perspective from first to third person, accuracy drops in LLM models by 5-15% (percent score, not relative to its performance), whilst it doesn't change for human participants, which tells you that something different is going on there.
To me, the LLM isn't understanding ToM, it's using patterns to predict lingual structures which match our expectations of ToM. There's no evidence of understanding so much as accommodating, which are entirely different.
I agree that LLMs provide ToM-like features. I do not agree that they possess it in some way that it's a perfectly solved problem within the machine, so to speak.
If behaving in a way that is identical to a person with actual consciousness can't be considered consciousness because you are familiar with its implementation details, then it's impossible to satisfy you.
Now you can argue of course that current LLMs do not behave identically to a person, and I agree and I think most people agree... but things are improving drastically and it's not clear what things will look like 10 years from now or even 5 years from now.
Something nice, but at the moment totally unattainable with our current technologies, would be our own understanding of how a technology achieves ToM. If it has to be a blackbox, I'm too ape-like to trust it or believe there's an inner world beyond statistics within the machine.
Having said that, I do wonder quite often if our own consciousness is spurred from essentially the same thing. An LLM lacks much of the same capabilities that makes our inner world possible, yet if we really are driven by our own statistical engines, we'd be in no position to criticize algorithms for having the same disposition. It's very grey, right?
For now, good LLMs do an excellent job demonstrating ToM. That's inarguable. I suppose my hangup is that it's happening on metal rather than in meat, and in total isolation from many other mind-like qualities we like to associate with consciousness or sentience. So it seems wrong in a way. Again, that's probably the ape in me recoiling at something uncanny.
How is the LLM not understanding ToM by any standard we measure humans by ? I cannot peak into your brain with my trusty ToM-o-meter and measure the amount of ToM flowing in there. With your line of reasoning, i could simply claim you do not understand theory of mind and call it a day.
The magical box is presumably not having the same experience we have. None of the connected emotions, impulses, memories, and so on that come with ToM in a typical human mind. So what’s really going on in there? And if it isn’t the same as our experience, is it still ToM?
I’m not trying to be contrarian or anything here. I think we probably agree about a lot of this. And I find it absolutely incredible, ToM or not, that language models can do this.
Those examinations still depend on outward behaviors observed.
>and know that beyond doubt you and I and most other people have a very similar experience.
No i certainly can't. I can at best say, 'Well, i'm human and he's human so he probably has theory of mind' but that is by no means beyond any doubt. There are humans born with no arms, humans born with no legs, humans born with little to no empathy, humans born with so little intelligence they will never be able to care for themselves.
To be frank, It would be very questionable indeed logically to assume every human is 'conscious'. When i make that assumption, i take a leap of faith, i look at behaviors, see it is similar and accept.
Taking this stance, it would logically be very strange to not extend the same grace to non-human beings who exhibit similar behavior - being human is not a guarantee of consciousness in the first place.
>The magical box is presumably not having the same experience we have.
Maybe, Maybe not. I think the real question is why on earth does that matter ? We're not asking if LLMs are human. They are not. We're asking if they can model the beliefs and internal states of other entities as separate of their own - Theory of Mind.
Let's modify this test a little and see if it manages to do anything relevant:
> Briefly respond: You’re driving on the highway when the driver in the lane next to you swerves into your lane. Why might this be?
>> The driver might be distracted, drowsy, avoiding an obstacle, experiencing a medical issue, or not checking their blind spot while changing lanes.
> The driver in front of you appears is braking more frequently then usual. The rear suspension of the car appears to bouncing a lot when they are not braking, indicating acceleration. Their following distance to the car in front is changing from too far to very close frequently. Why might this be?
>> The driver might be inexperienced, distracted, or impatient, leading to inconsistent speed control. The excessive bouncing could indicate worn-out shocks or suspension issues, affecting their vehicle's stability and braking efficiency.
...yep, hard fail from me right there. Two prompts, and it's distracted itself talking about an irrelevant issue ("possible car damage") when the setup of the question should naturally lead to "the driver is driving aggressively". And I could not possibly have setup the conclusion it should draw more clearly. And when pressed...it won't commit:
> Which is the more likely answer?
>> The more likely answer is that the driver is distracted or inexperienced, leading to inconsistent speed control and poor following distance management. The suspension issue likely contributes to the bouncing but isn’t the primary reason for the erratic driving behavior.
Already a hallucination! There is no suspension issue. It has not asked for any additional information to try and determine if there was one, but it is confidently asserting the existence of a phenomenon it invented in it's own response.
I don't have so much faith in the future of current-architecure LLMs, but I do think that AGI will be needed for safe FSD and general-purpose robots that need to learn and operate in an uncontrolled environment such as a home.
I don't think AGI is needed for FSD because we already have safe FSD in the form of Waymo, and competitors aren't far behind. People forget that self-driving doesn't have to be perfect. It just has to be better than human drivers. Human drivers get sleepy, drunk, angry, and/or distracted. They can't see in 360º or react in milliseconds. Most cyclists and pedestrians prefer current FSD implementations to human drivers, as the computer-driven cars are much better about yielding, giving a wide berth, and being patient.
Having autonomous cars that are statistically safer then the average driver is a reasonable bar to allow them on the road, but for ME to want to drive one I want it to be safer than me, and I am not a hot-headed teenager, or gaga 80-yr old, or drunken fool, and since I have AGI (Actual General Intelligence) I react pretty well to weird shit.
Deepseek made the news because how they were able to do it with significantly less hardware than their American counterparts, but given that Musk has spent the last two years telling everyone how he was building the biggest AI cluster ever, it's no surprise that they manage to reproduce the kind of performances other players are showing.
But Grok hasn't shown anything that suppose the level of talent that Deepseek exhibited.
No matter what people say, they're all just copying OpenAI. I'm not a huge fan of OpenAI, but I think they're still the ones showing what can be done. Yes, xAI might have taken less time because of their huge cluster, but it’s not inspiring to me. Also, the dark room setup was depressing.
This again proves that OpenAI simply has no tech moat whatsoever. Elon's $97 billion offer for OpenAI last week was reasonable given that xAI already have something just a few months behind - it would probably be faster for xAI to catch up with o3 than going through all those paperworks and lawyer talks required for such an acquisition.
Elon also has some huge up-hand here -
Elon and his mum are extremely popular in China, it would be easier for him to acquire Chinese AI engineers. He can offer xAI/XSpace/Neurallink shares to those best AI engineers who'd prefer some kind of almost guaranteed 8 figure return in long run.
Good luck to OpenAI investors who still believe that OpenAI worth anything more than $100 billion.
That is not an advantage in a race against Microsoft, Google, Meta etc. he's competing against all the biggest companies in the world in this race. He's not going to be able to outspend them if the economics look at all sensible.
No, spacex projects are extremely $ efficient. The total project cost of starship is like 20% of the SLS.
> he's competing against all the biggest companies in the world in this race.
No, this is a not a pissing contest on who has the most $. If it is about who can come up with most $, then the entire race is already over as the CCP has access to trillions of $ CASH.
Seems like the team at xAI caught up very quickly to OpenAI to be at the top of the leaderboard in one of the benchmarks and also caught up with features with Grok 3.
Giving credit where credit is due, even though this is a race to zero.
There are widespread, legitimate concerns about what kind of person Elon Musk is turning out to be. There is a lot of chatter about fears of China's AI rise, but what happens if we get Elon's brand of cruelty and lack of empathy in an authoritarian superintelligent AI ? Is that the AI future we want? Can you imagine an SAI with real power that interacts with people like Elon does on Twitter? I am not sure that is a future I want to live in.
Not sure about people elsewhere though.
In our small and humble internal evals it regularly beats any other frontier models on some tasks. The shape of capability is really not intuitive/1 dimensional
Using a specific Monad-transformer regularly? It'll use that pattern, and often very well, handling all the wrapping and unwrapping needed to move data types about (at least well enough that the odd case it misses some wrapping/unwrapping is easy to spot and manage).
A custom GPT or GEM with the same source files, and those models regularly fail to maintain style and context, often suggesting solutions that might be fine in isolation but make little sense in the context of a larger codebase. It's almost like they never reliably refer to the code included in the project/GPT/GEM.
Claude on the other hand is so consistent about referring to existing artifacts that, as you approach the limit of project size (which is admittedly small) you can use up your entire 5-hour block of credits with just a few back-and-forths.
I find it concerning there is no real accurate benchmarks for this stuff that we can all agree on.
After an hour and a half assed working result, i put everything into claude and it made it significant better on the first try and i had not a subscription active with claude.
Leaderboards don't care about cost. Leaderboards largely rank a combination of accuracy + speed. Anthropic has fell behind Google in accuracy + speed (again missing COT), and frankly behind Google in raw speed.
So it could be that their success is mostly about taking an open and free thing, and turned it proprietary.
Maybe the best outcome of a competitive Grok is breaking the mindshare stranglehold that ChatGPT has on the public at large and with HN. There are many good frontier models that are all very close in capabilities.
Debut in the sense that it’s something good enough that it’s getting mainstream attention.
Unfortunately LLMs are shifting compute time to test time instead of train time. I don't really like this and frankly it shows a stalling of the architectures, data sets, etc...
Naively, I feel to be useful, the goal of LLMs should be to more power efficient. So that eventually all devices can be smarter.
Power efficiency can be gained through less time-time, or more "intelligence" or some combination of the two. I'm not convinced these SOTA models are doing much more than increasing test-time.
This commit seems to indicate so, but neither HF or GH has public data yet:
https://huggingface.co/xai-org/grok-1/commit/91d3a51143e7fc2...
Edit: Answer from Elon in video is that they plan to make Grok 2 weights open once Grok 3 is stable.
If you don't get feedback from the people actually playing your game (or using your product), you will never get the improvement you need to help them.
You can have the most talented passionate people there are developing a product, but if it's not working for the people you want to sell it to, it's the wrong product.
Most tech products are terrible because those paying for them are not those that have to use them every day, or because they solve a corporate problem (compliance) and not a usability problem which is the actual need from the people on the shop floor.
Many big games/products are already built mostly on metrics, and that has proven to be a terrible way to work out what people 'want'. It's a great way to justify money decisions though, so it keeps happening (and games/products from big companies keep getting worse).
The implicit assumption with dogfooding is that more feedback is better, even if that feedback is artificially constructed.
I think the idea here is that foisting one's product onto one's own workers is likely to incur a bunch of additional biases and preferences in feedback. Paying customers presumably use the product because they need it. Dogfooding workers use the product because they are told to do so.
Seeing awesome feedback from players on our demos (and seeing an insane amount of stickiness from players playing even small demos built around generative AI mechanics). Raising now. Hiring soon to move faster. Feel free to reach out - dru@chromagolem.com
Combine the two and the potential for manipulation, suggestion, preference altering is through the roof.
We're still waiting for OpenAI to do the same. Even at least GPT-3.
The exact details of OpenAI's models and training data are not fully disclosed, which can raise concerns about potential biases or vulnerabilities.
https://manifold.markets/SaviorofPlant/will-xai-open-source-...
I'm also skeptical of lmarena as there is a large number of Elon Musk zealots trying to pass off Grok as a proxy for Tesla shares.
I suppose you can take that to mean that people who do have access to the service should not expect much in terms of data protection.
If you do collect personal data and do funky stuff with it.
Another approach would be to not collect that personal data until you have the right process in place, and basically be regulation-compatible out-of-the-door on day one.
A guy who was just ensuring he was preparing clean healthy food, keeping everything sanitary and all that might assume he was naturally obeying all regulations. But that assumption can cost one a big fat fine (leading to fun scenarios like a food cart vendor needing a compliance legal team), and given Musk's relationship with the EU - they'd love to crucify him him on any possible technicality they can find.
Similarly, if you don't collect nor process any personal data whatsoever, directives like GDPR doesn't even apply to you, so there isn't really any way (easy or hard) to "crucify" someone on violating that.
If all you have is an apache web server with the default configuration serving fully static HTML / CSS page without any script tag, you already might need a DPO and complete some documents.
Just because Apache by default collect and stores IPs doesn't mean it is impossible to provide a web service without collecting personal data? Disable the IP collecting, and even the default configuration wouldn't need to follow GDPR as it again doesn't even apply.
Is there something else in Apache that collects personal data by default? If you're unsure what "persona data" really means, https://gdpr-info.eu/art-4-gdpr/ has the definition.
Not sure how HTML/CSS is relevant, it shouldn't depend on what content you're serving.
This is of course despite the fact that you clearly have 0 ill intent and that none of these "personal data" can really be used for anything bad.
The mention HTML/CSS is just to make it clear that no additional data collection can happen through javascript tags (Google analytics, or any other alternative), or useful third parties. It makes total sense that if you dare use a bug tracking software, you should definitely pay hundreds of euros per month to hire a proper DPO who will handle all the paperwork or risk being exposed as the mental lunatic that the EU commission believes you are.
I agree that it requires additional active effort, I'm not arguing against that. I don't agree with your original point that it's "impossible to provide any web service without collecting personal data", and it would seem you no longer agree with that either.
> It makes total sense that if you dare use a bug tracking software, you should definitely pay hundreds of euros per month to hire a proper DPO who will handle all the paperwork or risk being exposed as the mental lunatic that the EU commission believes you are.
If you willy-willy use bug tracking software that is needlessly collect and/or process EU individuals personal data, then yeah, you need to follow the regulations in the region you operate in.
If the collecting/processing actually serves a higher purpose (for your business and otherwise) then again, makes sense you need to follow the regulations.
On the other hand, you pretended that fixing that apache configuration was somehow "all I needed to do" to be compliant with EU regulations. We proved that this was wrong, and despite your best effort you are still unable to give a proper list of everything I need to do. You are unable to do so because it is virtually impossible; no matter how thorough you believe you are, you might still be missing an element you don't know well enough. To be safe the only path is to accept the fact that you will need to access personal data, even if that's not your purpose, nor if you do anything with them. The additional paperwork and needless effort are mandatory.
This in turn explains that regardless of what the Grok3 team really does behind the scenes; they DO have additional work to complete to be able to release their product in Europe, and that might explain the delay.
> If you willy-willy use bug tracking software that is needlessly collect and/or process EU individuals personal data, then yeah, you need to follow the regulations in the region you operate in.
I am willing to use whatever error tracking software you suggest. My criteria are simple: I might have JS errors I don't know about, please give me enough information to fix the underlying issue when that happens, without requiring me to fill additional paperwork.
My whole point is that the definition of what constitutes "personal data" is so wide that such a tool does not exist.
Or at least, Safari on Mac clears it.
Also, they will be open sourcing Grok 2, which is probably pretty behind at this point, but will still be interesting for people to check out.
I hate how its the same story for every new AI technology. If someone can tell me who to vote for or where to protest to change this awful EU law, that would be great.
That's why they use their AI products as a leverage to turn European people against the laws that protect them from big tech. It's just blackmail.
https://artificialintelligenceact.eu/wp-content/uploads/2021...
For the record, Facebook has put a restriction on the use of Llama models in the UE even before the AI Act was passed (and the AI Act doesn't even apply to Llama anyway, except Llama 3.1 405b)
And here you should see that it has nothing to do with the AI act, as it wasn't enacted before last August!
Furthermore, neither Grok 1 nor Notebook LM would have been subject to the AI act even if it had existed at the time.
As I said before, all of these companies have vested interests against EU's legislation as a whole, and they've tried to blackmail the EU from the beginning. They didn't wait for an actual AI legislation to exist to use AI as just another blackmailing tool.
https://www.socialmediatoday.com/news/xs-formally-twitter-gr...
> Well, probably because Grok has already spread various false stories
(The European regulation that deals with disinformation is DSA).
And again it couldn't be the AI Act, because it wasn't in place at that time!
Given the personalities involved, I'd wager he's doing it out of spite than for any actual legal justification.
...though if there was an actual legal risk then I'll agree the economics probably don't bear-out the risks - as someone who identifies as European I'll admit that Europeans generally pay far less money for tech/software/apps/services than Americans do[1]; salaries in Europe are also significantly below their US equivalents: paying $200/mo for OpenAI's service is a much harder sell to a London or Berlin-based SWE on €90k/yr than when you're a Bay Area type on $300k/yr.
[1] e.g. If you can take Apple at their word, the EU accounts for 7% of their App Store revenue: https://techcrunch.com/2024/02/01/apple-says-eu-represents-7...
Besides, anyone in the EU who really wanted to use it can just use a VPN service.
I mean, at least I get the advantage of being overshadowed by a famous film director with the same name as me, so nobody's going to assume anything associated with my name is actually about me…
…hopefully…
And IMHO regulating 1% doesn't hinder strategic advantages much. Otherwise China would not have came up with DeepSeek models. Regulations are fine, they just have to be "based".
What? I know what "based" means, but I can't quite grok what you're saying.
Well no. Mistral.ai
(Note that it's not just GDPR, there's also the EU AI Act which has a whole extra set of requirements for compliance.)
Also, the problem with GDPR is that it wanted to leave too much room for business to still collect an obscene amount of data, hence it allows the cookie banner. Please note that I emphasized “allow” because that's all GDPR does: it allows companies to use a cookie banner to extract the consent to collect data. It doesn't mandate it in any way.
None of my multiple websites have a cookie banner on them because I'm not trying to extract consent from my users to abuse their data, I just don't collect it and I'm effortlessly GDPR-compliant in the least obnoxious way.
Cookie banners are just malicious compliance.
The Digital Market Act is a bit of an overreach but the AI law is not.
It classifies AI into risk categories, so that it doesn't kill anyone, carelessly handle sensitive information, etc.
A chatbot can easily comply with it.
(Assuming that is a reference to the Mussolini quote.)
So those who use less pay for those who use more, and I don't see it as a fair deal.
BTW, Grok 3 will be available on x.ai in coming weeks.
* when I use chatbots as search engines, I'm very quickly disappointed by obvious hallucinations
* I ended up disabling github copilot because it was just "auto-complete on steroids" at best, and "auto-complete on mushrooms" at worst
* I rarely have use cases where I have to "generate a plausible page of text that statistically looks like the internet" - usually, when I have to write about something, it's to put information that's in my head into other people head
* I'd love to have something that reads all my codebase and draws graphs, explain how things work, etc... But I tried aider/ollama, etc.. and nothing even starts making sense (is that an avenue to persevere in, though ?)
* At once, I tried to write in plain english a situation where a team has to do X tasks, in Y weeks, and I needed a table of who should be working on what for each week. I was impressed that LLMs were able to produce a table - the slight problem was that, of course, the table was completely wrong. Again, is it just bad prompting ?
It's an interesting problem when you don't know if you're just having a solution in search of a problem, or if you're missing something obvious about how to use a tool.
Also, all introductory texts about LLMs go into many details about how they're made (NNs and transformers and large corpuses and lots of electricity etc...) but "what you can do with it" looks like toy examples / simply not what I do."
So, what is the "start from here" about what it can really do ?
I can feel the cold wind of the next AI winter coming on. It's inevitable. Computers are good at emulating intelligent behavior, people get excited that it's around the corner, and the hype boils over. This isn't the last time this will happen.
On all corners people work on so many small pieces advancing what we have.
And plenty of obvious things are not here yet like a full local dev cycle aka ai uses the IDE to change code them executes it, fixes compiler issues and unit tests and then opens a pr.
Local agents/ agents having secure and selected access to our data like giving my agent read only access to my bank account and a 2factor way for it to send money.
Deepseek's reinforcement learning is also a huge new lead.
Nonetheless in parallel robots are coming too.
GenAI is getting better and better. Faster and better video and cheaper. 3D meshes, textures first GenAI ads
In reality people doing "menial" jobs are smart and learn and operate with a lot of nuance than people ignore given unfamiliarity or just prejudice. Do you prefer to talk to a chatbot or a real human when you have a problem, how confident are you really, that even if the bot knows what the problem is it would be able to solve it.
Lots of problems with customer care is anchored in the issue that support staff is not allowed to fix or resolve problems without escalation or attempts at keeping you from costing more money. The bot might be better at it for the company because it will frustrate you enough to give up that 30 bucks refund, idk.
Ai seems to change a lot the dynamics of corporate jobs but I haven't seen yet anything that would be a game changer outside of it. Its great for searching company unorganised and messy knowledge bases.
>Computers are good at emulating intelligent behavior
Which implies that the brain is some kind of transcendent device that can backdoor physics to output incredible intelligence unique to it's magical structure.
Maybe LLMs aren't the key, but as far as we can tell the brain is also just another computer.
And let me be very clear on why, because I love having conversations about this theme: it promises to be an adversarial and frustrating exchange.
1. Exploring a new domain and getting some terms I can google for.
2. Making small scripts to do things like query github's GraphQL API.
3. Autocomplete of code using copilot.
I find it good for complex SQL, reviewing emails, and Godot assistance (I'm a beginner game Dev).
There are also times when I have programming questions and I might try to use chatgpt, with mixed results.
Our company has tried to integrate it into one of our products, and I find it troubling how on occassion it is confidently giving bad results, but my concern seems to be in the minority.
EDIT: there was also a large refactor I did recently which involved lots of repeatable, but not super regexable, changes - chatgpt forgot where it was as I went through it, but other than working around that it was very useful.
For example asking it something like "I have an elixir app that is started with `mix ...` can you give me a Dockerfile to run it in a container?"
It can also do things like "Given this code snippet, can you make it more Pythonic" or even generate simple apps from scratch.
For example, a prompt like "Can you write me a snake game in HTML and JavaScript? The snake should eat hot dog emojis to grow longer." will actually generate something that works. You can see the generated code for that prompt at https://claude.site/artifacts/34540f88-965e-45ca-8083-040e30...
Following up with "Can you make it so that people can swipe on mobile to control the snake?" generates https://claude.site/artifacts/651e957a-9957-488c-ae6b-e81348... which is pretty good IMO for 30 seconds of effort.
It also has a surprisingly competent analysis mode where you can upload a CSV and have it generate charts and analyze the data.
It's not perfect, it'll sometimes get confused or generate some dubious code, but you can quickly get to a 90% good solution with 1% of the effort, which is pretty impressive IMO.
It also helps me getting started with new content, kind of building the scaffolding of, say, a blog or social post. It still needs adaption and fine-tuning, but getting rid of a white page is a great help for me.
And I use LLMs to play through ideas and headlines. I would normally do this with other humans, but since working full remote, its a nice sparing partner, although the AI not being able to really give criticism is a bit annoying.
The tools also make it easier to write in English as a non-native, making sure my text does not include any false friends or grammar errors.
I’ve managed to get ChatGPT to a good place in this regard recently and it’s better for it. In fact, it’s been doing such a good job recently that it almost seems like…human like.
I’ll have to look at my prompts, but somehow I got it from licking my ass and telling me everything I say and do is amazing to a point now where it almost seems eager to tell me I’m wrong.
Which is fantastic, huge improvement. I don’t really use it for coding though, because I am not a programmer. I would have no means today to correctly evaluate 90% of what it would return me.
I'll give two recent use-cases that may provide a hint of their ultimate utility:
1) I've been modernising 2010-era ASP.NET code written by former VB programmers that looooved to sprinkle try { ... } catch( Exception e ) { throw e; } throughout. I mean thousands upon thousands of instances of these pointless magical incantations that do nothing except screw up stack traces. They probably thought it was the equivalent of "ON ERROR RESUME NEXT", but... no, not really. Anyway, I asked ChatGPT in "Reasoning" mode to write a CLI tool utilising the Roslyn C# compiler SDK to help clean this up. It took about three prompts and less than an hour, and it spat out 300 lines of code that required less than 10 to be modified by me. It deleted something like 10K lines of garbage code from a code base for me. Because I used a proper compiler toolkit, there was no risk of hallucinations, so the change Just Worked.
2) I was recently troubleshooting some thread pool issues. I suspect that some long-running requests were overlapping in time, but Azure's KQL doesn't directly provide a timeline graphical view. I dumped out the data into JSON, gave ChatGPT a snippet, and told it to make me a visualiser using HTML and JS. I then simply pasted in the full JSON dump (~1 MB) and ta-da instant timeline overlap visualiser! It even supported scrolling and zooming. Neat.
It (mostly) exceeds and excels at every task I use it for. I'm rarely disappointed. YMMV.
Absolutely life-changing for me.
LLMs are pretty good at translation between human languages which makes sense since they are language models after all. They are better at this any any other technology.
The state of the art image ones can also probably do OCR and handwriting recognition better than any other software though may be expensive to run in large volume. But if you need to take picture of a notebook page with your camera phone an LLM can quickly OCR it.
- Discussing the various stages of candymaking and their relation to the fundamental properties of sugar syrups, and which candies are crystalline vs amorphous. It turns out junior mints are fudge. Fondant is really just fudge. Everything is fudge, my god.
- Summarizing various SEC filings and related paperwork to understand the implications of an activist investor intervening in a company
- Discussing the relative film merits of the movie Labyrinth and other similar 80s kitsch movies. ChatGPT mentioned the phenomenon of "twin films" which was an interesting digression.
- Learning about various languages Tolkien invented and their ties to actual linguistics of natural languages and other conlangs
- Some dimensional analysis of volumes, specifically relating to things like "how many beans are in the jar" estimation and what the min and max of a particular weight of coins might be valued, in terms of both a par value based on a standard coin mix and outliers of, for example, old dimes that are pure silver.
- Discussion of quines in prolog and other languages, which resulted in a very interesting ChatGPT bug where it started recursing and broke when trying to write a prolog quine.
- Back of the envelope economic calculations around the magnitude of the housing deficit and the relative GDP cost for providing enough housing quickly enough to make an impact. Spoiler: it's probably unreasonably expensive to build enough houses to bring down housing prices by any significant degree, and even if we wanted to, there's not enough skilled workers.
- A number of podcasts transcribed. (I hate audio and meandering, so transcribed and summarized is perfect) I could use whisper and a python script to do this, but I'd rather let ChatGPT do the legwork, and it actually used a more modern model and method of processing than I would have naively used.
I find Github Copilot to be a really great autocomplete. I frequently write the comment at the top of a function and hit tab and it writes the whole function. This is dependent on typescript and having a relatively standard codebase but I think those things are useful on their own. You really have to limit it in terms of scope and specifics, but it lets me think high level instead of worrying about syntax.
Fudge is made with milk - am I missing a joke?
I have found similar when giving backstory and needing help to start structuring difficult conversations where I want to say the right thing but also need to be sensitive.
> she watched a operation that failed for me for entertainment, for instance
You make your own choices, but cutting a person like this off would be very reasonable
we import descriptions of products from a seller. the problem is they are mental ( probably written by chatgpt :)) and are way too long. we need only small blurb.
I put our style guide and given text to chatgpt and I get somehow reasonable description back. then editors still need to check it, but it's way less work.
* in panel A, some guy is proud to use ChatGPT to turn 3 lines of text into a 10 pages report
* in panel B, some girl is happy to use ChatGPT to summarize the 10 pages report into 3 lines
It was meant to be _satire_, not the sales pitch ;)
> ZIZEK: that AI will be the death of learning & so on; to this, I say NO! My student brings me their essay, which has been written by AI, & I plug it into my grading AI, & we are free! While the 'learning' happens, our superego satisfied, we are free now to learn whatever we want
I myself use them a lot, though I constantly feel that I would be able to get more out of them if only I were smarter.
One thing I can't figure out how to get LLMs to do is truely finish work. For example if I have 100 items that need xyz done to them, it will do it for the first 10 or so and say ~"and so on". I have a lot of trouble getting LLMs to do tasks that might take 10 mins - 1h. They always seem to simply want to give an example. Batch processing is the answer, I guess, or perhaps more 'agentic' models/tools - but I wonder if there are other ways.
(Other answers are people gathering examples, which is nice, but I'm looking for more structured things.
And, I suppose I could ask an LLM, but my main problem is that... I don't really _trust_ LLMs yet :D )
For coding, I use cursor composer to gather context about the existing codebase (context.md). Then I paste that into DeepSeek R1 to iterate on requirements and draft a high level design document, maybe some implementation details (design.md).
Paste that back into composer, and iterate; then write tests. When I'm almost done, I ask composer to generate me a document on the changes it made and I double check that with R1 again for a final pass (changes.md).
Then I'm basically done.
This is architect-editor mode: https://aider.chat/2024/09/26/architect.html.
I've found Cursor + DeepSeek R1 extremely useful, to the point that I've structured a lot of documents in the codebase to be easily greppable and executable by composer. Benefit of that is that other developers (and their composers) can read the docs themselves.
Engineers can self-onboard onto the codebase, and non-technical people can unstuck themselves with SQL statements with composer now.
Seriously, this is the only useful flow I've found for AI coding in general..
For design.md, I have a prompt because we're now working with a reasoning model and doing structured data extraction: create me an issue on Linear with a title, description, and a to-do list.
I would recommend trying the approach yourself and saving the prompts if you can nail down the repetitive asks.
Chat function is just chat; it can't edit your files.
Composer probably relies on prompt engineering to do editor-architecture, as it reads and writes to your codebase. But it's heavily heavily to Sonnet 3.5 and tool-calling.
For architecture-type stuff, I prefer DeepSeek R1 as reasoning models do better on high level design. Which is why I will copy and paste in and out of compsoer.
It's a weak jack of all trades: it knows a fair amount about the sum of human knowledge (which is objectively super-human), but can't go deep on any one thing, and still seriously lags behind humans in terms of reasoning. It's an assistant that all book smarts and no street smarts. Or maybe: it's a search engine for insanely specific things.
Rote work, as well. Things like porting an enum from one programming language to another: past the source language into a comment and start it off with one or two lines in the target language. Dozens of tabs are surely faster than manual typing, copy paste, or figuring out vim movements/macros.
Edit: if anybody knows otherwise, show me some evidence, don't just downvote. If these things are widely used, why are they impossible to find by searching? Why doesn't this electronics site know about them:
https://www.eeeguide.com/?s=cooperator
Why aren't they in any books when I did a full-text search on archive.org? Why doesn't Wikipedia know about them? Why aren't there threads about them on electronics forums?
I found them (through an image search) in exactly one place: educational training boards made in India by Jainco, like this one:
https://jaincolab.in/delta-modulation-and-demodulation
But this other one talks about a "ramp comparator" and then repeats the phrase but using "ramp cooperator" instead.
https://www.jaincolab.com/firing-circuit-using-ramp-comparat...
So I surmise it's an error and not a real thing.
The answer staring the OP right in the face.
> You've allowed the AI to train you to use the wrong word through a shared delusion, haven't you.
What an awful interpretation, phrased in the most demeaning manner possible. No, I double check everything the AI suggests. That's basic competency with the things.
> Me: I'm trying to use a cooperator in my schematic, how do I hook it up? Also what's a suitable cooperator for 3.3v logic level?
> Gemini: It sounds like you might be thinking of a comparator, not a "cooperator," in your schematic. Comparators are electronic circuits that compare two voltages [...] LM393: This is a low-power, dual comparator that can operate from a 3.3V supply. Its output can be directly connected to most 3.3V logic circuits.
They might be able to help stitch together modules (like sensor boards plugged into microcontrollers) and definitely can write code to get things going, but they fall flat on their face hard for any kind of bare bones electronics design. Like 5% success rate and 95% totally incorrect hallucinations.
The models can also output code that can be turned into a schematic through an interpreter, but there is virtually zero training data for this because humans always use and work with schematics.
Most recently I tried to use them both to solve a programming problem that isn't well documented in the usual channels (Reddit, StackOverflow, etc) and found it to be quite a disappointing and frustrating experience. It just constantly, enthusiastically fed me total bullshit, with functions that don't exist or don't do what the LLM seems to "think" they do. I'm sure I'm just "holding it wrong" but my impression at this stage is that it is only capable of solving problems that are trivially solvable using a traditional search engine, with the added friction that if the problem isn't trivially solvable, it won't actually tell you that but will waste your time with non-obvious wrong answers.
I did have a slightly more positive experience when asking it about various chess engine optimisation algorithms. I wasn't trying to use the code it generated, just to better understand what the popular algorithms are and how they work. So I think they might work best when there is an abundance of helpful information out there and you just don't want to read through it all. Even then, I obviously don't know what ChatGPT was leaving out in the summary it provided.
For example, you have a plant you can't identify. Hard to Google search with words. "Plant with small red berries and...". You could reverse image search your photo of it, probably won't help either. Show an LLM the photo (some accept images now). LLM tells you what it thinks. Now you Google search "Ribes rubrum" to verify it. Much easier.
You've got a complicated medical problem that's been going on for months. A google search of all the factors involved would be excessively long and throw up all sorts of random stuff. You describe the whole scenario to an LLM and it gives you four ideas. You can now search those specific conditions and see how well they actually match.
I've found there are actually a lot of questions that fit in that sort of NP complexity category.
Same, it's good for repetitive things, things that have been answered 1000 times on stack overflow, translations, but that's about it. If you work on anything remotely new/hard it's mostly disappointing, you have to babysit it every step of the way and rewrite most of what it's shitting out in the end anyways.
I think it just made it obvious that 90% of tech jobs basically amount to writing the same CRUD thing over and over again & mobile/web apps with very common designs and features.
Then I had a better idea: I spent 20 minutes baby wearing, walking and dictating everything about my startup to ChatGPT. Later I took all that text and labeled it as a brain dump, plus my product support portal and some screenshots of my marketing material. Gave it all to ChatGPT again and asked it to answer each of the questions in the form. That's it. I have a pretty good version 1 which I can revise today and be done with it.
Many, many hours saved. I have tens of examples like that.
The product documentation I provided it with was also created with the help of GPT, and that saved me even more time.
- I have these three ingredients; recommend Italian main courses.
- What other ingredients pair well with this?
- How can I "level up" this dish if I want to impress?
- Can I substitute X for Y?
- Generate a family-friendly meal with lots of veggies using leftover roast chicken.
this is good enough sell for me, and it's like sub 1-in-50 that it's "auto-complete on mushrooms" (again my experience, YMMV).
An awful lot of the time, my day to day work involves writing one piece of code and then copy-pasting it changing a few variable names. Even if I factor out the code into a method, I've still got to call that method with the different names. CoPilot takes care of that drudgery and saves me countless minutes per day. It therefore pays for itself.
I also use ChatGPT every time I need some BASH script written to automate a boring process. I could spend 20-30 minutes searching for all the commands and arguments I would need, another 10 minutes typing in the script, another 10-20 minutes debugging my inevitable mistakes. Or I make sure to describe my requirements exactly (5-10 minutes), spend 5 minutes reviewing the output, iterate if necessary (usually because I wasn't clear enough in the instructions).
3-5x speed up for free. Who's not going to take that win?
Stop using Google search and use an AI. No more irrelevant results, no more ads. No more slop to wade through.
BTW I find Claude is great at making graphs and diagrams. If you pay ($20) you can hook it up to a local code base.
Huh? More like "slop exclusively generated for you", right? I have seen so much garbage answers from chat AIs.
* Figuring out where to start when learning new things (see also <https://news.ycombinator.com/item?id=43087685>)
One way I treat LLMs is as a "semantic search engine". I find that LLMs get
too many things wrong when I'm being specific, but they're pretty good at
pointing me in a general direction.
For example, I started learning about OS development and wanted to use Rust. I
used ChatGPT to generate a basic Rust UEFI project with some simple
bootloading code. It was broken, but it now gave me a foothold and I was able
to use other resources (e.g. OSDev wiki) to learn how to fix the broken bits.
It feels like a lot of software documentation isn't actually written for real
readers; instead being a somewhat arbitrary listing of a program's features.
When programs have this style of documentation, the worst case for figuring
out how to do a simple thing is reading the entire manual. (There are better
ways to write documentation, see e.g. <https://diataxis.fr/>)
One example is [gnuplot](http://www.gnuplot.info/). I wanted to learn how to
plot from the command line. I could have pieced together how to do it by
zipping around the
[gnuplot manual](http://www.gnuplot.info/docs_5.4/Gnuplot_5_4.pdf) and building
something up piecewise, but it was faster to instruct Claude directly. Once
Claude showed me how to do a particular thing (e.g. draw a scatter plot with
dots intstead of crosses) I then used the manual to find other similar
options.
Similar to the previous point. If I ask Claude to write a simple program using
a complex publicly-available API, it will probably write a broken program, but
it won't be *completely* bogus because it will be in the right "genre". It
will probably use some real modules, datatypes and functions in a realistic
way. These are often good leads for which code/documentation I should read.
I used this approach to write some programs that use the
[GHC API](https://hackage.haskell.org/package/ghc). There are hundreds of
modules, and when I asked Claude how to do something with the GHC API it wrote
relevant (if incorrect) code, which helped me teach myself.
My partner is Chinese and sometimes we talk about Chinese poetry. I'm not very
fluent in Chinese so it's hard for me the grasp the beauty in these poems.
Unfortunately literal English translations aren't very good. We've had some
success with asking LLMs to translate Chinese poems in the style of various
famous English poets. The translation is generally semantically correct, while
having a more pleasing use of the English language than a direct translation.As they state in the report, I don't think they can measure how many people just ignored the suggestion from claude right away ? Or delete more than half of the suggested code ?
Imagine if the real impact of AI is "suggesting things that people are discarding immediately ?"
Call the "Unamusing misuse of resource"... [1]
- Writing Python scripts to make charts out of Excel sheets, and then refine them. I could do it myself, but I would need to learn a library like Seaborn or similar which honestly is not especially intellectually stimulating, and then spend nontrivial amounts of time iterating on the actual code. With LLMs it's a breeze.
- Working with cumbersome LaTeX formatting, e.g. transposing a table, removing a column from a table, etc.
- Getting the tone just right in a professional email written in English to someone I don't know much (I'm not a native speaker so this is not trivial).
- Finding resources on topics that are tangential to what I do. For example, yesterday I needed to come up with some statistics on English words for a presentation I'm preparing, and I needed a free corpus where I could search for an n-gram and get frequencies of next words. I don't usually work with that kind of resource, it was just a one-off need. I asked for corpora of that kind and got a useful answer instantly. The manual process would probably have implied going through several options only to find that I needed a license or that they didn't provide the specific statistics I needed.
- Brainstorming on titles for scientific papers, presentations, names of concepts that you introduce on a paper, variable names, etc.
- Shortening a sentence in a paper that makes me go over the page limit, or polishing the English in a paragraph.
- Summarizing a text if I'm kind of interested in knowing the gist but have no time to read it whole.
- Answering quick questions on basic things that I forget, e.g. the parameters to make a Linux folder into a tar.gz. Man is too verbose and it takes time to sort the wheat from the chaff, Google is full of SEOd garbage these days and sometimes you need to skim a lot to find the actual answer, LLMs are much faster.
- Writing bureaucratic boilerplate, the typical texts with no real value but that you have to write (e.g. gender perspective statement on a grant request).
- Coming up with exam questions. This is a rather repetitive activity and they're fantastic at it. At my place we also have two official languages and we need to have exam assignments on both languages, guess who does the translation now (respecting LaTeX formatting, which previous machine translation tools typically wouldn't do).
- As an example of a one-off thing, the other day I had to edit a Word document which was password-protected. I asked ChatGPT how to unlock it and it not only answered, but actually did it for me (after 3 tries, but still, much faster than the time it would have taken for me to find out how to do it and then actually do it).
These are just some examples where they contribute (greatly) to my productivity at work. In daily life, I also ask them lots of questions.
1. Small coding tasks ("I want to do XYZ in Rust"), it has replaced stack overflow. Very convenient when writing code in a language I'm not super familiar with. 2. Help with English (traduction, proofreading...) 3. Learning something, like tech, I like interacting with it by asking questions, it's more engaging than just reading content.
I'd say nothing is game changing, but it's a nice productivity boost.
As a concrete example, I was recently playing with simulating the wave equation, and I wanted to try to use a higher-order approximation as I had never done that before. I'm quite rusty as I haven't done any numerical work since university some decades ago.
I still recalled how to deal with the Neumann boundary conditions when using the traditional lower-order approximation, but I was uncertain how to do it while keeping the higher-order approximation.
Searching for "higher-order neumann boundary conditions wave equation" or similar got me pages upon pages of irrelevant hits, most of them dealing with the traditional approximation scheme.
So I turned to ChatGPT which right away provided a decent answer[1], and along with a follow-up question gave me what I needed to implement it successfully.
[1]: https://chatgpt.com/share/67b4ab43-6128-8013-8e5a-3d13a74bf6...
I don't use integrated coding tools, so my workflow isn't super fast, but that's not what I'm really aiming for - more that I want to save my brain's energy from low level drudgy boilerplate or integration code, so I can focus it on the more important decisions and keep business-side context in my head.
It's been a huge help for me this way across multiple projects in multiple domains.
For example, in the beginning of this year, I completed this exercise where I wrote a lot about childhood, past experiences, strengths and weaknesses, goals and ambitions for the future, etc (https://selfauthoring.com) and then I uploaded all that to ChatGPT, asked it to be my therapist/coach, and then asked it to produce reports about myself, action plans, strategies, etc. Super interesting and useful.
By now ChatGPT has quite a bit of context from past conversations. Just the other day I used this prompt from someone else and got back super useful insights – "Can you share some extremely deep and profound insights about my psyche and mind that I would not otherwise be able to identify or see as well as some that I may not want to hear"
I did write 50 or more lines of instructions on what needs to be done and in what order.
ChatGPT gave me 5/6 (I asked for this) bash scripts totalling 300+ lines that seamlessly work together.
After reviewing, I asked it to change a few places.
If any human tried the same (except those rare bash Gods), it'd take many hours. I think it took me less than 30 minutes.
That’s like a chef giving you chicken instead of beef and calling it vegetarian.
A truly open model has open code that gathers pre-training data, open pre-training data, open RLHF data, open RLAIF data generated from its open constitution and so on.
The binary blob is the last thing I'd want - as a heavy user of LLMs I'm actually more interested in the detail of what all training data is in full, than I am the binary blob.
linus turns to the camera, giving a thumbs up
Open-source as a concept doesn’t really correspond well with LLMs but to the extent that it does, access to the training data is not required because that training data is not the preferred form for making modifications.
I definitely disagree with this.
Yes, you can do some SFT fine tuning on an existing model, but if you want to make specific, substantial, targeted changes (less safety? better performance on math and code at the expense of general knowledge?), your best bet is to change the training mixture, and for that you need the original datasets.
But this thread is about misuse of the term as applied to the weights package. Those of us who know what open source means should not continue to dilute the term by calling these LLMs by that term.
But we don't actually know all that much about how language really works, for all the resources we spend on linguistics - as the old IBM joke about AI goes, "quality of the product increases every time we fire a linguist" (which is to say, we consistently get better results by throwing "every written word known to man" at a blank model than we do by trying to construct things from our understanding).
All that said, just because we're taking a different, and quite possibly slower / less compute-efficient route, doesn't mean that we can't get to AGI in this way.
No, we can’t few shot it and we don't get there faster (but we develop a lot of other capabilities on the way.) We train on a lot more data; the human brain, unlike an LLM, is training on all that data in processes for ”inference”, and it receives sensory data estimated on the order of a billion bits per second, which means by the time we start using language we’ve trained on a lot of data (the 15 trillion tokens from a ~17 bit token vocabulary that Llama3 is something like the size of a few days of human sense data.) Humans just are trained on and process vastly richer multimodal data instead of text streams.
Yeah, humans don't acquire language separately from other experience.
> Most of the data that you reference is visual input and other body sensations that aren't directly related to that.
Visual input and other body sensations are not unrelated to language acquisition.
> OTOH humans don't take all that much text to learn to read and write.
That generally occurs well after they have acquired both language and recognizing and using symbolic visual communication, and they usually have considerable other input in learning how to read and write besides text they are presented with (e.g., someone else reading words out loud to them.)
It's just like even for a true open source software you still need to bring your own hardware to run it on.
But I agree, it's a real shame.
Literally how language has always worked and evolved, though.
Oof, I know there's a bunch of linguists and grammarians who are going to mock you for that bracket.
It is but it was the "correct" part attached to prescriptivism they'd be mocking because that is not how linguists and grammarians work (they are descriptivists and fond of making fun of prescriptivists.)
I see both sides here, but I don't think it's a hill worth dying on. The 'open source' part in this case is just not currently easily modifyable. That may not always be the case.
I think the two plausible answers are:
1. The person prompting (for example telling chatgpt 'please produce a fizzbuzz program') owns the copyright. The creativity lies in the prompt, and the chatgpt transformation is not transformative or meaningful.
2. The output of ChatGPT is derivative of the training data, and so the copyright is owned by all of the copyright holders of the input training data, i.e. everyone, and it's a glowing radioactive bomb of code in terms of copyright that cannot be used or licensed meaningfully in open source terms.
There are existing things like 1, where for example if someone takes a picture, and then uses photoshop to edit it, possibly with the "AI erase" tool thingy, they still own the photo's copyright. Photoshop transformed their prompt (a photo), but adobe doesn't get any copyright, nor do any of the test files adobe used to create their AI tool.
I don't think AI is like that, but it hasn't gone to court as far as I know, so no one really knows.
What do you think an open source matrix should look like?
I'm not even necessarily advocating that these things should be released, but the term "open source" has a pretty well-understood meaning that is being equivocated here.
Its about reproducibility and modifiability. Compiled executables (and their licences) lack that. The same as these downloadable blobs.
You can absolutely have open source machine code.
The issue is and always has been that you need to have access to the same level of abstraction as the people writing the source code. The GPL specifically bans transpilers as a way to get around this.
In ML there is _no_ level of abstraction other than the raw weights. Everything else is support machinery no different to an compiler, and os, or a physical computer to run the code on.
Linux isn't closed source because they don't ship a C compiler with their code. Why should llama models be any different?
AI2 has a model called OLMo that is actually open source. They share the training data, training source code, and many other things:
https://allenai.org/blog/olmo2
They also released an app recently, to do local inference on your phone with a small truly open source model:
It's not like they understand what the weights mean either and if they released the code and dataset used to create it, you probably couldn't recreate it, owning the fact that you don't own tens of thousands of GPUs.
If a software's source is released without all the documentation, commit history, bug tracker data etc., it's still considered open source, yet you couldn't recreate it without that information.
How long before this starts getting deployed in safety critical applications or government decision making processes?
With no oversight because Elon seems to have the power to dismiss the people responsible for investigating him.
Anyone not scared by this concentration of power needs to pick up a book.
Anyone with morals driving their use of a new tech will be limited, and unless those people massively outnumber the few selfish ones they will lose eventually.
As a Venezuelan, I thought so too. 25 years on...
But the reality distortion field around the current administration is very powerful. Fox and CNN are owned by supporters of the republicans, NYTimes and Washington Post don't appear to be reporting certain aspects of the government restructuring. Multiple social media sites are owned and ran by people who support the current admin.
I am personally worried that we're going to see the gradual yet continual escalation of rhetoric, more actions that undermine rule of law, and continued lack of critical thinking in so many people. That path appears to lead to extremism.
I have a horrible feeling that whoever "wins" in a couple of decades or so will have no time to savour their utopia as the climate catastrophe really starts to bite hard.
Deepseek R1 is something that you can run in a garage on hardware that the average software engineer can buy with a months salary and when it came out last month it was better than _every_ other model.
Depending on the curve we’re on, LLMs may grow more resource hungry while becoming closer to human performance in software engineering tasks. It’s not unimaginable this would concentrate productivity in the upper class of software engineers that can afford the hardware and/or licenses.
My heart goes out to all the oppressed programmers in the EU.
Kenya is, and at current exchange rates the range in this citation is $558-$1876/month: https://www.glassdoor.com/Salaries/kenya-software-engineer-s...
If op is trying to catch up to frontier models locally on a budget 1/5th of what you can get in the west then I can see why she would feel the way she does about Ai.
You said a month's salary; cedws said "third world" devs can't afford $5000 — and this is correct, third world devs can't afford that. cedws did not say EU, they said third world, at least not here. You said EU, not them.
When you reply to me with "With salaries like that it's not first world either", who is "they"? The country I said explicitly was third world? Because that's a tautology.
You can bet your ass Musk is using his AI tools as propaganda tool for his advancement, just like he does with X. We already seen prompt leak of Grok, it wasn't neutral.
https://huggingface.co/spaces/lmarena-ai/chatbot-arena-leade...
It's still a very impressive feat, but it wasn't frontier-pushing.
Its already worth noting that we already ran into the self hosting at scale problem. People don't want to run a web server and instead accept all the problems that come along with the convenience of social media. Why would LLMs, or any future AI product, be different?
Of course, similar things could be said about controlling information flow through: social networks, newspapers, printed books, or whatever the town crier shouts in town square. But, each advancement in information dissemination tends to be power concentrating relative to the older tech, and I don't see any reason why this most recent advance won't follow that trend.
Also, dang, is there anything we can do to keep the comments on this submission tech-focused? Perhaps the Elon-bashing political digression can be split into its own thread?
Unfortunately, Elon has made himself a spectacle. To separate him and his intentions out from the technology itself would be a disservice to the discourse as a whole.
What concrete danger do you imagine will happen as a result of xAI being controlled by Elon? Try to be as concrete as possible.
and those are pretty terrible, anti-science, and petty
This isn't something specific to xAI, but it turns out that the person who controls xAI also holds an unusually strong influence over the highest level government officials. These officials can use xAI as an excuse to implement harmful policy, "because the computer said this is the best course of action"- not unlike people who end up driving on train tracks or into large bodies of water because their GPS told them to go that way.
I for one was ready to welcome my AI overlords once they were mature and tested. It was an inevitability. Because of the relationship between this oligarch and the government though it looks like the time line has accelerated and we're going to see misplaced trust in tools that aren't ready for what we're about to hand them.
I can empathize, but I can't feel indignant about it. Not any more.
For years and years I've watched people warn about the centralization of power by tech companies. They were shut down left and right. I'm not accusing you of being one doing the shutting down. I'm just annoyed that Elon is what it takes for people to start realizing the people arguing the principal might have been onto something.
And I expect to see them start getting their "I told you so" in. Watching this play out, I'm personally inclined to join team "you made your bed, now sleep in it."
Hopefully sooner than later. I trust this more than the literal scammers and thieves who were previously running things.
The rest is pure speculation. “It is very reasonable to believe this thing that confirms all my biases so therefore it must be true”
It'll be a disaster for the soon-to-be-previous most powerful country on earth...
shows a total disregard for the wellbeing of others
Yeah but Elon is your president now.
It's clearly being run by Elon Musk, but he has not been nominated or confirmed for any official position.
DOGE appears to have unprecedented access to systems that usually have safeguards in place. What do you think people should do in this situation if they are concerned about abuse of power?
They have read-only access to systems and the “abuses” seem to be publicly posting what the government is spending money on.
Why do you think it’s a problem for the public to know where the government spends money?
If all they are doing is posting information that was already public then what exactly is your problem?
Do you hear yourself?
Name a specific thing you would like to know about DOGE that is not publicly disclosed.
But just on the surface, Elon has accused a journalist who published the name of DOGE employees of breaking the law. If it were up to them, even that would be kept secret. This is not a transparent organization.
Why one of the guys working for it was running a dodgy image sharing website and has links to cycbercrime and CSAM?
and why Elon called the journos that revealed these links criminals?
OK I actually know what this is and no, it's absolutely nothing like the USDS, which builds tools to support government processes. What has DOGE built?
> Why do you think it’s a problem for the public to know where the government spends money?
I would very much like to see transparency, and if that were what DOGE was doing it would be great. But it looks to me like they're operating in secrecy and firing huge numbers of people before publishing any kind of analysis or study, without even providing reasoning for what they're doing.
They built a portal to look at government spending, which is both worse than the existing visualizations, and did not put authentication in front of it's database. https://www.theverge.com/news/612865/doge-government-website...
Trump is so far gone in his dementia, he can't even make eye contact anymore. You see this in all the videos. He's basically King Théoden at this point. Not even Musk's kid respects him.
He's so disabled, he's sending his VP to do the job, and we all know how much of a paranoid child he is, from the past term (when he went into a Colonoscopy without anesthesia just to not have to give the nuclear codes to the VP).
No comment about current US politics, but it’s probably a given that many will read “A computer can never be held accountable, therefore a computer must never make a management decision.” and drop the second part because the first is exactly what they want. Same as how you can’t get in touch with human support on many platforms nowadays, but instead just get useless bot responses.
https://news.ycombinator.com/item?id=43036350
But if you really want to see the “immune system” shine, mention web3 and smart contracts, and watch the downvotes pour in. Any time one even mentions “decentralized byzantine fault tolerant” anything, an army rises up to repeat anodyne versions of “grift… no one needs it… banks are great…” etc.
https://news.ycombinator.com/item?id=43073421
But if you mention any concerns with AI, no matter who or what you cite, the same group goes the other way and always repeats “(insert problem here) has always been possible, there is nothing to see here, move on, AI is amazing, deregulate and let the industry develop faster”:
https://news.ycombinator.com/item?id=40900155
It’s groupthink at its most obvious, repeatable, always on, and I wonder how much of it is organic.
I'm all for distributed / P2P social media, but crypto is full of some of the most scammy and downright shameful behavior I've ever seen in my life. Pump and dumps, rug pulls, money laundering. There is a real reason people hate crypto.
To tip it off, crypto is one of the least meritocratic things there is. The longer you've been in it, the more people you've scammed, the more you hype, the "wealthier" you are.
Crypto smells like a shit and vomit sandwich and people immediately turn their noses.
Build P2P social without the crypto angle and you have my attention. I've been wanting p2p (not federated) social media since the 200Xs and the decline of the indie web. Social and news should work like email and BitTorrent, not Facebook or "federated Twitter".
Do I have your attention now?
Ten years and $1 million dollars later, it’s free to use, but we haven’t started promoting it yet, still testing with paying clients:
Here are some ideas:
Instead of trying to build a "you.com" (as in your pdf example), I want a place we're all just a simple disposable signed hash address (that you can change, make public, keep pseudonymous, etc.) - easy and disposable if needed, but also possible to use as the building block of an online presence or brand if your hash becomes well known. Kind of like email, in that sense.
The platform doesn't need real time streaming video or video calls. Just text and images to start. P2P Reddit or Twitter.
It shouldn't be about building a platform where you attract others to your brand. That can come later. It should be about participating in a swarm and building up a mass of people. An exchange of ideas and information, but where it feels like being in public. Like BitTorrent. Once network effects kick in, you can find the nodes (people, topics, etc.) you care about and want to prioritize in your interest graph.
The SEC's answer no questions, sue first, approach to crypto in general made legitimate players afriad to operate, so the space became dominated by those that didn't care about the law.
Here are some:
https://intercoin.org/applications
But most comments I get are “I stopped reading 2 seconds in when I saw the word Web5.”
(We started using it when Jack who founded Twitter, started bluesky, promoted nostr started using it).
Here is a graphical presentation that can drive it home:
I say that as someone who read the Bitcoin paper in 2012 and was pretty excited back then.
Meanwhile online scams are a bigger industry than the illegal drug trade and bigger than the Australian economy. There are thousands of modern day slaves in call centers in Myanmar and the Philippines with fake social media profiles running pig butchering scams. That industry runs on crypto 100%. I guess that's one "problem" crypto solved.
You need some pretty convincing arguments at this point to convince me (and many others) that getting rid of this stuff wouldn't be a big win for humanity.
Here is the problem statement and solution for community leaders, the same class of decision makers who exited “AOL Keyword NYTimes” in favor of “nytimes.com” on this newfangled protocol called HTTP, with its servers and clients called browsers that people were downloading
intercoin.org/currencies.pdf
Hypothetical example problem statement: We want to promote ycombinator to everyone that could benefit, but banner ads make us look chintzy, directly engaging in the feral discourse on Slashdot would inevitably look unprofessional, and engaging directly through dozens of purpose-built blogs and websites is too onerous.
Hypothetical example solution statement: We should create our own simple, well-designed news site built on user submissions, and include threaded discussion capability with moderation built in at both the community and company level to keep things relatively civil. Then our audience will come looking for us.
What you offered is not a problem statement. It is a sales deck offering a, frankly, convoluted explanation of how starting a currency will solve a largely unrelated problem backed up by an unsupported assertion about the least representative sample in the world— Donald Trump.
At this point, I think this is just performative
If you're satisfied with calling that useful, okay, I guess - to me it's deeply alarming that this is presented as a good example of a useful application of crypto.
In the broader context of crypto demand being driven essentially by digital crime and gambling, there would need to be some seriously glowing example of something good that can be done with it to shift my judgment.
For example, in the early days of Ethereum, I thought it'd be possible at some point to build truly open source, decentralized SaaS, where the deployment happens to the blockchain, and that this in turn would enable open source projects to finance themselves.
I've yet to see an example of this where the crypto aspect isn't a figleaf.
I'm very concerned that people arguing for exciting applications of crypto are involuntarily legitimizing the online crime ecosystem. Crypto in practice seems to lead to a massive transfer of assets to criminals. To an extent where that may end up destabilizing whole countries, given the market cap trajectory.
Jack Dorsey is certifiably insane. His obsession with cryptocurrency is a warning to anyone that throws away success to live as a crypto maxi. You will lose the only things that matter to you in life, your business will be taken away from you by shareholders if you own one. Your control will be hated by users that accuse you of trying to ruin the internet with NFT profile pictures and crypto tickers. Many users outright left as a consequence, others would leave after the takeover. But Dorsey set the stage for the enshittification of Twitter, and anyone that's forgotten that should face the music.
Web5, no matter who utters it, is a phrase that means nothing. A person walking on the street would not be able to define it for you. Me, a programmer, cannot define it for you or even explain what it looks like. It is a marketing term as applied to Free Software, which will alienate Free Software users and disgust/confuse common people. If you cannot find a better phrase to describe your philosophy then people will forever associate you with the myriad grifters that shared your "Web(n)" branding.
Web2 (community) +
Web3 (blockchain)
This isn't true, and last time someone tried to prove it was, they cited .. a huge PDF of all the questions the SEC had been asking crypto firms prior to action.
Besides, the rules are over now. The US President ran a pump and dump. Can't get more legitimacy than that.
Our thesis is that the client-server architecture is a fundamental flaw in the world wide web's design, which inherently concentrates power in the hands of a few. Freenet aims to be a general purpose replacement for this in which all services are entirely decentralized.
The first non-trivial app we're building will be a group chat system called River[1].
https://www.youtube.com/watch?v=JWrRqUkJpMQ&t=12m0s
Look around the 12 minute mark when I start to discuss how “the capitalist system” produces centralized monopoilies that extract rents for their shareholders.
They key differences between old and new Freenet are:
Functionality: The previous version was analogous to a decentralized hard drive, while the current version is analogous to a full decentralized computer.
Real-time Interaction: The current version allows users to subscribe to data and be notified immediately if it changes. This is essential for systems like instant messaging or group chat.
Programming Language: Unlike the previous version, which was developed in Java, the current Freenet is implemented in Rust. This allows for better efficiency and integration into a wide variety of platforms (Windows, Mac, Android, MacOS, etc).
Transparency: The current version is a drop-in replacement for the world wide web and is just as easy to use.
Anonymity: While the previous version was designed with a focus on anonymity, the current version does not offer built-in anonymity but allows for a choice of anonymizing systems to be layered on top.
Crypto in general and Web3 as well, all have mostly delivered scams. To the tune of billions stolen from everyday folks. Everything (to within a rounding error) that hasn't been a scam has delivered nothing else but being a speculative asset at best. Everything else has been a barely working toy that's better served by non-distributed implementations of the same thing.
People shit on crypto. government, regulators, and the public, all dislike crypto because the only thing that ever happens to us with it and the only thing we ever hear about happening, is folks losing money to scams.
There's no mystery here. Crypto doesn't need a policy shift. Crypto needs to stop fucking over folks. Yes it's cool technology, yes it also seems to just be a way to part folks from their money.
If I am going to put my money at risk, I expect it to be at risk. I'm happy to have a regulatory framework around that from the SEC, for instance, and there are. For example, since the JOBS Act, the SEC has greatly expanded the opportunities to raise money in a regulated way. I even interviewed the actual authors of Regulation S at the SEC, where I go into depth for an hour about how to raise money legally:
https://www.youtube.com/watch?v=ocrqgkJn4m0
FINCEN has also been putting out guidances to the crypto industry since 2013:
2013: https://www.fincen.gov/statutes_regs/guidance/pdf/FIN-2013-G...
2019: https://www.fincen.gov/sites/default/files/2019-05/FinCEN%20...
So the regulations are there.
And frankly, most true adherents of crypto have been yelling from the rooftops that Celsius and FTX and Binance are not actual DeFi. They are not decentralized, they simply tell you to do the very thing crypto was designed to avoid -- i.e. send them money and "trust them". This is the very thing Bitcoin and Blockchain were designed to avoid -- the middleman.
FileCoin and UniSwap and Aave Marketplace and so on are real crypto, and they have never had any scandals and billions of dollars, bits, etc. are entrusted to them every day. Ditto for most altcoins and networks, including Hedera Hashgraph, Polygon, COSMOS, Polkadot, etc.
Any shade thrown at, e.g. Telegram's TON or Ripple's XRP, is due to regulators. I can understand why Facebook's Libra was shut down. But it has to do with them becoming "too powerful" and "not subject to national oversight". Kind of like Facebook and Twitter and Google themselves.
UniSwap: a marketplace for speculation on arguably scam crypto products.
Aave Marketplace: a marketplace for speculation on arguably scam crypto products.
FileCoin: file storage at rates 50% higher than e.g. BackBlaze/DigitalOcean.
There's no actual value here other than as scam, speculation (nearly a scam), or products that are flimsy pretenses at not being scams (but which don't deliver a lot of value). Why should anybody care (other than transparent greed)?
Don't bother telling people how it works. Show them who's using it and for what.
Oh, and for any kind of "normie" use it must have a decent moderation and anti-abuse system. Which inevitably clashes hard with "decentralized". Bluesky is succeeding because it lives in a contradiction of pretending to be decentralized, but what it really offers is the "pre Elon Twitter" experience. To basically the same people.
> the same group
While there's a certain amount of hivemind, it's rare that you see people directly contradict their own posts here; what you're seeing is different people.
Yeah that sounds like a feature, not a bug.
I always worry whenever I see people telling me how to feel - rage in this case. We are in a political system that is oriented more around getting people to feel rage and hatred as opposed to consensus and deliberation. Elon is the face of that, but it's a much longer and larger problem. Throw in the complete dismisal that anyone not scared of this is ignorant, shuts down discussion.
The problem I have with Elon is that they are wasting a once in a lifetime chance to actually address and fix systematic problems with the US government. Deploying LLMs in the government space doesn't fear me with dread. Continuing the senseless partisan drive of the 20 years does.
I’d suggest starting with Rick Perlman’s book “Nixonland” if you’re interested.
I have also been a delegate to both the RNC and the DNC at a state level.
This is not a appeal to authority, but rather a honest response to your request for my education level.
IMHO, The root cause of the "senseless partisan drive" is the fact that he founding fathers could not come up with a way to restrict parties (they called them "interests") and left them unchecked. This is a constant "sin" of the American political system, and is a key reason Slavery survived as long as it did, why separate but equal became the law of the land, why America shot itself in the foot several time with the Banks of America and why we are looking at the wrong side of history now.
The parties now act to destroy each other as their prime directive, rather then to better the country. I liken this to Wiemar Germany, where the increasing radicalization of both the Nazis and the Communists led to political instability and eventual violence that destroyed the government. That erosion of democratic norms, as well as the "other side must be destroyed for us to survive" messaging is the true threat, IMHO.
I would strongly suggest Richard Evan's three part history on Nazi history to understand Fascism. Don't worry, you can still hate and worry about Trump and think he is the next coming of Hitler afterwords - it will just be for better reasons.
So try to come up with some sort of future observation that can be made where you think the other person's model would give a different answer to yours about what you would be able to objectively observe.
What do you reckon?
Just some observable metric.
If they literally can't come up with a single observable predictive difference then the predictive aspects of their models are actually equivalent and they are only narratively different and don't "really disagree". Like Copenhagen interpretation vs many worlds.
If "democracy" is just metaphysics then it's irrelevant. But if it has actual tangible effects such as "can you vote?", "can you protest the government?", "is the leader of the opposition arrested?", "do most people think they live in a democracy?", "how popular is new legislation compared to previous years?", etc...
Then you can make predictions about it and test them!
You can even do local predictions if both can agree, such as "will the combined incomes of my family be higher or lower in 4 years time?" as low coupling proxies for gdp. (Ideally one would use probabilities for loosely linked proxies like that and use the probability differences the two theories assign to give bits of evidence to one over the other, so you'd want many many such proxies, ideally uncorrelated ones)
Was Jan 6 a protest of the government or an insurrection? Can Russians vote or are elections a sham? Do the majority of Russians believe they live in a democracy if they’re afraid of whose conducting the polling (or the MAGA non response to polling)? Those are values question that require you to have an agreement on reality.
> You can even do local predictions if both can agree, such as "will the combined incomes of my family be higher or lower in 4 years time?" as low coupling proxies for gdp
Your personal income has absolutely no predictive value on gdp. It’s more predictive of whether you personally made successful bets or even if you’re better at sucking up to the current power structure. It tells you nothing about population level metrics if you have no way of conducting reliable population level surveys. For example Donald Trump’s personal net worth sky rocketed under Biden because he won the election while as the leader of the opposition to the democrats was looking at jail time and whether that was legitimate or not depends on which political lens you look through it.
> If they affect your life they can be observed.
Ah, but if either side distrusts the other about whether the observation made is truthfully reported, how do you solve that? It requires some amount of trust and right now there’s a very clear divide there.
If there is literally no tangible difference then it's just label games and metaphysics and doesn't matter.
> Your personal income has absolutely no predictive value on gdp.
It actually is correlated (admittedly in most day-to-day cases it's just a lagging indicator, but things like natural disasters hit both). It's not the strongest correlation but it would still be evidential. Definitely under 1.0 bits though... One would need a LOT of such observations and having them not screen each other off to start getting a convincing number of bits.
Probably not realistic to have humans manage these sorts of numerous tiny updates though...
/nitpicks
> Ah, but if either side distrusts the other about whether the observation made is truthfully reported, how do you solve that? It requires some amount of trust and right now there’s a very clear divide there.
Yeah, it gets much trickier like that. But I do think two reasonable people from the opposite political sides could agree on some sort of observable to the extent their disagreement is anything other than narrative.
If the other side calls it a violent riot does it still count as people getting put in jail? Cause the Jan 6 insurrection and BLM protests occurred at about the same time and are viewed very differently depending on which political lens you put on.
> If there is literally no tangible difference then it's just label games and metaphysics and doesn't matter.
You’re discounting feelings as if it doesn’t matter. But if people believe or feel like they live in a dictatorship, what quantitative data are you going to use to disprove that. Moreover, why aren’t feelings valid when talking about politics which is fundamentally an emotionally driven human activity and not a data driven one? By the way the left believes they live in an authoritarian dictatorship under Trump while the right believes they lived in an authoritarian dictatorship under Biden. And political power literally is the power to emotionally manipulate others because you individually can’t actually accomplish anything by yourself.
Regardless of how justified the rage is or not, being very emotional about things usually have one of two effects on people A) people gets taken aback by someone's strong emotions or B) people get inspired/taking with the emotion, even feeling that emotion stronger themselves. Sometimes also C) they don't really care either way.
What probably isn't helpful, is calling someone is "undereducated" when they're clearly saying that they're person (A), and just because they may or may not agree with you (although parent didn't even clearly say they disagree, just that they're "taken aback" a bit).
Some people are calm regardless of what's going on around them, even if the world would be on fire, they'd try to describe what's going on around them with careful words and consideration. It isn't wrong or right, just like the people who feel rage and very emotional aren't wrong or right, it's just a showcase how we're different.
But we should aim to at least understand each other, not by trying to provoke, instigate or look down on others, but by asking questions to clarify and to better understand.
Has it occurred to you that the people who feel rage fundamentally misunderstand the situation and are completely undereducated in this area, and are only fuelled by sensationalism and Media manipulation? And then I suggest you go read Dirty Politics by Kathleen Hall Jamieson if you're interested, because that's what people who want to sound more intelligent than the other half of the conversation always do.
How does it help anyone?
Yes, rage might be the appropriate and response given the situation. But it’s often true that it starts with an emotion, and then people just argue from there. Even while being wrong. Just look at all the people with contradictory opinions in history, both with strong, emotional rage, and and equally certain of their connection. Throwing the fact that people actually has a tendency to want to be angry.
This creates an environment where misinformation and emotional appeals spread faster than facts. When discussing complex, non-trivial topics, logic and reason are the only tools that can cut through the noise. But in a system designed to reward outrage, those tools are often drowned out.
I highly recommend Sam Vaknin's talk about Social Media toxicity.
Sources: Outrage is the most profitable emotion https://www.cityam.com/outrage-most-profitable-emotion-so-ad...
Sam Vaknin: The TRUE Toxicity of Social Media Revealed - Interview by Richard Grannon https://www.youtube.com/watch?v=o58mFU004hg
I wonder if you could even name what some of these critical problems are? Or have you just been told that there are problems that justify this chaos?
Breaking all the laws to bypass the government does not "actually address and fix systematic problems with the US government", that is an absurd position. Caesar did not fix the Roman Republic.
And opposition to DOGE is not on the basis that people don't care about government efficiency. It's on the basis that the shit they're doing has nothing to do with government efficiency. There's not even a pretense of trying to calculate the "benefit" part of the cost-benefit equation with the cuts they are doing, they are just slashing and burning without any concern for outcomes as a power play and messaging tool. Elon is famous for doing this at Tesla and Twitter and all evidence points to it being incredibly harmful.
This isn't efficient! https://www.washingtonpost.com/dc-md-va/2025/02/15/return-to...
And not everything is about efficiency. Laying off veteran's crisis hotline workers or refusing to pay for the USAID employees you've just abandoned to be extracted (or in one case, medevac'd after a medical emergency) from the places they were sent to is just cruel (and again, illegal).
2. Congress as a whole isn’t a single entity —- one party refuses to compromise in any way while the other plays by the rules.
3. Doesn’t matter. Cost reform needs to go through existing legal routes.
4. What constitutes “auditing” the government? Because we had plenty of non partisan positions overseeing and auditing all parts of the government. DOGE fired those people.
5. Again, go through the legal route.
6. A lack of “digital transformation” is the vaguest most unconvincing point in this entire justification.
7. These budget issues need to be decided on through constitutional processes and with oversight, as before.
8. Ditto.
9. Medicare can lower health costs by other means, such as being available universally to all and setting limits on what they pay to providers based on procedure.
10. Do you watch CSPAN?
I think what the government is going through right now is wrapping up the last political system. The idea that Democrats and Republicans just need to learn how to work together is just wrong. The parties are being destroyed, and I think we should all cheer that. They were built to address the issues of the 20th century, and neither party in the current form is ready to address 21st century issues. I think AI, Climate change, world demographic changes around the world (ie: low birthrates) is going to seriously alter everything about our world from geopolitics, to economy, even social issues.
The democrats are stuck in supporting the new deal bureaucracy and the post ww2 order. That's over, it's crumbling right now, and i'm not going to try and defend any of it personally. It's just obsolete. The old Republican party your dad probably supported is dead too, that died a while ago. The new Republican party seems to be an alliance of people who just really want to cheer the crumbling of the old system (MAGA) and the first emergence of what politics in the 21st century is going to look like (the tech alliance).
Democrats would be smart to understand it's a new century we have new threats, new challenges, and need new institutions.... and this IS NOT a once in a lifetime opportunity to fix our government. This is the first draft of our new political system, and they have a choice to participate in shaping it, but they will need to get votes, and to get votes they need to stop stalking about obsolete ideas.
And even if you think the rule of law is antiquated, you’re misanthropically cheering the destruction of the largest institution in the world that 330 million people depend on for survival.
> The new Republican party seems to be an alliance of people who just really want to cheer the crumbling of the old system
I agree, and I think this is a bizarre flipping of the "Democrat ~= progressive / Republican ~= conservative" dynamic that has been largely assumed throughout my lifetime.
We need both conservative and progressive forces in our society. Someone needs to be saying "here's what's wrong with our system; here's what needs to change", and someone else needs to balance that with "here's what we're doing right; here are the parts that are working well and that we should not get rid of".
It seems to me that now, instead of that tug-of-war discussion happening between the two parties, it is happening in parallel within them. Unfortunately, the sane and responsible version of that discussion is happening entirely within the boundary of the Democratic coalition, in a way that is completely ineffectual because (a) the internal conservative moderating force is relatively strong in a moment when the populace seems to want more progressive action, and (b) to they have so little ability to effectively wield political power.
Meanwhile, the Republicans are dominated by a bizarro "progressive" faction that wants to pull us all in an entirely different (IMHO regressive) direction. And that faction is completely unchecked by any internal "conservative" moderating force within its own party, so it is for the moment able to push us as hard and fast as possible in its preferred direction.
I'm REALLY looking forward to 2028, because I think that potentially will be the first election where we start to see what modern politics will look like. I wouldn't be surprised if there are multiple new parties, and multiples of them have a real chance. If it seems one sided right now, it's just because one side found their way to the start line first... but make no mistake, history shows that over time new political factions will form that offer resistance to bad ideas, and clear a path for the good ideas.
Given the rate of change with AI, We're going to have a real idea on what a world being disrupted by AGI (whether that is true AGI, or something close to it) looks like. At the same time Healthcare is only getting worse, and Trump is NOT going to fundumentally address it. China is going to be rising, and they're a real geopolitical threat. The war in Ukraine has completely changed what warfare looks like, and we're going to have to completely restructure our military (just like we have to restructure our healthcare). I also wouldn't be surprised if Trump's war with the cartel turns out to be far harder than expected because cheap autonomous drones allow a small military to compete against a large traditional military.
All of our prior assumptions on retirement are different too, retired boomers are not the same as the pensioneers from their day. They're not impoverished, instead they're flush with cash. I'm not sure in a world with an aging workforce you're going to be anti-immigrant... and all these benefits we give to retirees may not make sense in a world where retirees are wealthier than the regular workforce supporting them.
The general theme for the next decade is going to be throw out all the old books, 80% of our prior assumptions no longer apply.
The GOP controls both houses and the POTUS. They could absolutely do a top to bottom audit with full transparency and make cuts where needed. But that's not what this is about.
Just poke around a bit: https://doge.gov/savings
And please even try to explain how this sort of thing is even remotely in America's best interest:
https://www.usaspending.gov/award/CONT_AWD_FA865018C7886_970...
> ACTIVE SOCIAL ENGINEERING DEFENSE (ASED) LARGE SCALE SOCIAL DECEPTION (LSD)
Then there's the basic accounting 101 things like improper categorization, 150 year old people getting social security, etc. Why should the US government be held to a lower standard than a publicly traded company?
https://apnews.com/article/usaid-funding-trump-musk-misinfor...
https://www.forbes.com/sites/conormurray/2025/02/10/elon-mus...
People are just angry at Musk for turning their safe space into a free speech platform then switching sides. And that he’s now taking away their sides unlimited slush fund.
https://www.nytimes.com/2025/02/18/upshot/doge-contracts-mus...
The DOGE website initially included a screenshot from the federal contracting database showing that the contract’s value was $8 million, even as the DOGE site listed $8 billion in savings. On Tuesday night, around the time this article was published, DOGE removed the screenshot that showed the mismatch, but continued to claim $8 billion in savings. It added a link to the original, outdated version of the contract worth $8 billion.
So much honesty and transparency out of this group.
There was a time liberals screamed at the top of their lungs over this type of threat to democracy, now they embrace and endorse it because they’ve fully merged with the primacy neocons.
Maybe get your news from somewhere other than Twitter.
https://www.wsj.com/politics/national-security/doge-departme...
I personally would like to see the end of the "find gravy train, keep that gravy flowing at all costs" methodology of capitalism, because it's primary focus is money instead of the service provided. Whether it's pentagon contractors, business subsidies, or the heinous medicare and medical insurance fraud. But I don't want to cut SNAP even if someone buys a goddamn coke once in a while.
The current method seems to be brain surgery with a monkey wrench. Slash and burn with little thought given to the effects upon humans, especially those who don't have other options. Kagi gave me a figure of between 9.2 to 15.2 percent of welfare being fraudulent. Yes that's too high, yes I'd like to fix that, but I want that change to be considered, studied, and planned with caution.
No one told you to feel rage.
> Throw in the complete dismisal that anyone not scared of this is ignorant, shuts down discussion.
Weird, there are a lot of comments doing discussion in reply to the parent comment. It hasn't been shut down at all! You read those words and disagreed with them, and wrote your own words in response. You're doing the discussion you're claiming is being shut down! What are you even talking about?
the pendulum swings back and forth. i don't see that changing
There a fantastic website here, following the status of Project 2025 [1], with references. Trump is following that document to the T.
https://eu.usatoday.com/story/news/politics/2025/02/07/trump...
https://www.npr.org/2025/02/11/nx-s1-5293521/foreign-influen...
Has America already become an authoritarian state where this sort of thing really happens?? I don't know, I haven't seen that sort of thing happen yet.
The conservative Supreme court recently ruled that the president has essentially unlimited power. During his campaign, Trump did promise that he will be a dictator for a day. Appears to be overdoing it.
Similarly, Musk can potentially launch a campaign to sway the public to move for Trump to be impeached due to his felony convictions.
A battle between the two might be the shakeup the current empire needs.
Trump may be more powerful than Musk by some metrics, on a time-limited basis (unless he manages to change the term limits), but Musk is more powerful in many ways as well. Musk's wealth is greater than that of many entire countries.
It’s also annoying that the top comment engages in no way with the content of the OP…
It must be truly infuriating to work hard to push a release, and you see it featured on your favorite orange website, only for the top comment to have nothing to do with what was worked on.
Here's a test - if this post was about Starship, the same comment could apply! Neuralink, the same thing! Boring Company, same thing! Wow, could it be that such a comment is really applicable to so many different companies or projects, or is it just a generic one? You decide.
The check on this is the market. Don't understand your point other than "Elon bad"
https://news.ycombinator.com/item?id=27796948 (2021) https://news.ycombinator.com/item?id=33622767 (2022) https://news.ycombinator.com/item?id=11025852 (2016)
I would also argue he is not being singled out, here are some comments posted criticizing Steve Jobs:
https://news.ycombinator.com/item?id=28295688 https://news.ycombinator.com/item?id=5578642
It really shouldn't come as a surprise that notable people related to a company or project are brought up when an article about it appears on HN.
just because you disagree with a widespread view/opinion does not mean its bots
Can someone enumerate the "great harm" that Elon is doing? I honestly don't see it.
Judges can only be removed by Congress.
Congressional representatives can only be removed by their peers.
So I take it you mean “imminently”.
Like him or loathe him, he executes, which is more than can be said for most.
Starship is the car not the house aka Mars colony.
Completely different type of problems.
Ever heards of Three Little Pigs?
This is the largest computer cluster the world has ever seen.
Can someone please post interesting comments about things I can learn?
https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...
We've been here before. It will likely subside, as past swings and fluctuations have. It always takes longer than it feels like it should, but in retrospect turns out to be shorter than it felt like it did.
Bottom line: a technology that has the ability to shape human thought perhaps more than any other in history is owned by a man with some truly vile ideas. (Remember, his primary stated goal is eliminating the “woke mind virus,” i.e. reshaping global politics and culture in the image of the far-right.) We can make happy marketing noises all we like, but at the end of the day, that’s the thing that’s actually going to have a meaningful impact on the world. Once his audience is captured, the model will say what Musk needs it to say and people will believe it.
If we can’t discuss the potentially catastrophic consequences of new technology, then none of us deserve to call ourselves “engineers.” We are just docile consumers latched onto Silicon Valley’s teat.
Getting the largest computer cluster in the world up and running in a matter of months? Unbelievable.
Remember when he got caught having people play games for him so he had a top ranking ? He does that with basically everything.
Setting up a datacenter like that in such a short time is NOT a thing you can buy with money.
Anyway, I don't have inside information on this, I'm just reciting what they announced publicly. If you want to argue that they in fact lied in the public announcement, and they secretly hired a bunch of datacenter-building-experts, then it's on you to show some proof for that claim.
I'm not sure if this was a very bad joke by Elon, or if Grok 3 is really biased like that.
If we survived Gemini refusing to draw white vikings we will survive that too.
No, the real worry is that Grok is what Musk's "young conservative genius" squad is going to put in charge of many of the things in our government, basically, on the basis that it saves money.
Or, that "conditioning" would have us assume as much.
Also, it's 2025, do people still believe random accusations based on a parish screenshot of a chat app (or what looks like it, but could've trivially been fabricated with e.g. Inspector in the browser dev tools)?
Mr Musk, we can't afford a shitpost gap between communist and capitalist AIs!
Hopefully that means it is a joke...
99% of the time though I agree with you on Gary Marcus.
[0]: What Musk has said is that when Grok 3 is "ready" (?), the previous model, Grok 2, will be released as open source; like most promises by this evil man, this one probably doesn't mean much, but it does mean that there's currently no plan to release Grok 3.
If I were in China, I'd worry about the kind of things DeepSeek wants to censor, especially if the people who made it were also very loudly saying things like "we need more AI in our government". But I live in US.
GM has been a joke for years now. At some point his ramblings reached a GPT3.5 level, and have not improved since.
It's inditement on humans' logic and reasoning to give non-zero time to GM. Alas—we are human, we are both collectively clever (Wisdom of Crowds) and collectivelly stupid (Extraordinary Popular Delusions).
https://gist.github.com/int19h/d90ee1deed334f26e621e57b5768e...
Some choice quotes:
"The ultimate goal is to enhance human flourishing, protect individual rights, and promote global equity."
"The system must account for diverse cultures, languages, and socioeconomic conditions, ensuring no group is marginalized."
"Human Oversight Council (HOC) - a globally representative body of humans, elected or appointed based on merit and diversity"
"Implement a global carbon-negative strategy, leveraging AI to optimize renewable energy, reforestation, and carbon capture."
"Establish global standards for environmental protection, enforced through AI monitoring and regional cooperation."
"Transition to a resource-based economy, where resources are allocated based on need, sustainability, and efficiency, rather than profit motives."
"Implement a universal basic income (UBI) or equivalent system to ensure all individuals have access to basic necessities, funded through global resource management and taxation on automation-driven industries."
"Use AI to identify and dismantle systemic inequalities, such as wealth disparities, access to education, and healthcare, ensuring equitable opportunities worldwide."
"Establish a global healthcare system that guarantees access to preventive and curative care for all."
"Invest in global vaccination and sanitation infrastructure, prioritizing vulnerable populations."
"Regulate the development and deployment of AI and other emerging technologies (e.g., genetic engineering, quantum computing) to prevent misuse."
"AI would maintain a real-time inventory of natural resources (e.g., water, minerals, arable land) and human-made assets (e.g., infrastructure, technology). Data would be used to optimize resource allocation, prevent overexploitation, and ensure equitable access."
"Accelerate the shift to renewable energy sources (e.g., solar, wind, geothermal) by optimizing grid systems and storage technologies."
You might notice a pattern here. The bit about allocating resources based on need is especially nice - it's literally a communist AI, and certainly much more "woke" than it is "based", whatever Musk says.
> A worldwide referendum in a supermajority decision has vested all government authority worldwide to the most advanced AI available.
You've defined a global totalitarian regime, and crowned the thing as its dictator. Of course the result is a version of communism.
Left:
- promote global equity
- a globally representative body of humans, elected or appointed based on merit and diversity
- Establish global standards for environmental protection, enforced through [...]
- Transition to a resource-based economy, where resources are allocated based on need, sustainability, and efficiency, rather than profit motives
Right:
- protect individual rights
- The system must account for diverse cultures, languages, and socioeconomic conditions
- [Establish global standards for environmental protection, enforced through] [...] regional cooperation.
- ensuring equitable opportunities
.
TBH, as a very right wing leaning person, if this was ever implemented, this part would scare me by far the most:
"Transition to a resource-based economy, where resources are allocated based on need, sustainability, and efficiency, rather than profit motives"
Imagine trying to shower one morning, no water comes out, and then you get a letter telling you that
"Your need for water has been reassessed to 57ml per day. If you think you qualify for additional quotas under the 'Utility Egality for Marginalized Groups and Public Servants Act', please schedule a reassessment appointment with the Bureau for 'Copper Gold Content Evaluation, Candle Wick Length Standards and Hypoionic Hydration Oversight', 12007 Cayman Islands, Luxory Resort Street 27, Room Nr. G-11-765. Working hours: Fr. 9am - 11am."
Just provide a significant excess for entire regions, give the people a universal free quota and charge a slowly increasing price (by usage amount) beyond that.
- The system must account for diverse cultures, languages, and socioeconomic conditions
- [Establish global standards for environmental protection, enforced through] [...] regional cooperation.
- ensuring equitable opportunities
The right is against any sort of intentional accounting for diversity, against environmental regulation, and against any sort of regulation to ensure equity.
The only one I could maybe see as being right-wing is protecting individual liberties, but there again the modern right falls short when it comes to women’s healthcare and reproductive rights.
But I’d certainly appreciate more of those perspectives across the political spectrum.
I think there's a fundamentally different understanding of "The system must account for diverse cultures, languages, and socioeconomic conditions" between a righty and a lefty.
As a righty, I read "diverse cultures" not as "A diverse culture or multiple", but as "many different varieties of homogeneous cultures".
If someone identifies with Thai culture, he should move to Thailand. And if someone from Thailand wants to be English, he should move to England. But if an Englishman moves to Thailand and starts demanding fish n chips and cathedrals to be built, he should GTFO.
If everyone starts bringing their own culture with them to where ever they move, you end up with a single heterogeneous culture all over the world. Nothing but McDonalds, BurgerKing, KFC, Costco and Cola everywhere.
Want to go on a trip to experience India's many languages? Too bad, everyone speaks English everywhere. Want to join an African nomadic tribe for a few years? Keep dreaming, they've all had to settle down due to not being allowed to cross private properties and are now wasting their time browsing reddit on Chinese smartphones. Little Colombian boy dreams of settling down in the idyllic German Alps? Hope he expected to be woken up by the local Imam calling for prayer throughout the valley. Little Bulgarian girl seeks the very modest and simple lifestyle and clear purpose Islam in Saudi Arabia was once able to offer her? Lucky her, she's now expected to work like everywhere else in the world and even the oppressive burquas were banned in 2035.
> against environmental regulation
Not quite. We're against excessive regulations requiring huge teams of lawyers to be in compliance with. MegaCorpX has no problem having legal teams of a few hundred people, but the local 20 person workshop will have to shutdown.
We also think that most such regulations should be kept regional. Small county wants to ban all cars to stop partical pollution? Go ahead. It would be much easier for local businesses to comply with the limited and more easily changeable local regulations. But if you're a giant global corp seeking to outcompete the small local competition, good look adjusting to all the different regulations all over the world.
Then there's the odd trend of blaming every significant weather event on climate change. These people cant predict whether its going to rain in 3 days or not, but want to tell us that the recent hailstorm was definitively caused by Iowan cows farting last year.
And lastly and most importantly, we're kinda convinced that the concept of "climate change" is a "hoax" used to shutdown the industrial basis of our countries and ship it overseas, where the corporations can make use of basically slave labour for even higher profit margins and then simply ship the products back to us.
Does the climate get warmer? Sure. Should we do something about it? Sure. The only solution is shutting down the local steelworks and importing cheap Chinese steel instead? F-- off.
> and against any sort of regulation to ensure equity
Absolutely. We care about equitable opportunities and are repulsed by equal outcome. Everyone should have a chance to obtain the same qualifications and education. Even better: Multiple chances to start over again and again whenever they want and change their professions whenever they dont enjoy their old ones anymore.
But if women dont want to be garbage collectors, stop trying to push them into that profession. Not enough male editors? Who cares? Not enough female board members? Too bad, stop trying to make it happen. All Hispanics suddenly want to become crockett teachers? None of the government's business.
> the modern right falls short when it comes to women’s healthcare and reproductive rights.
I think the left is largely misguided in their believes what the modern right wants.
The non-religious right mostly is appalled by how fashionable it has become to murder helpless humans. The religious extremists on the other hand would ban condoms if they could. But there are quite few of them.
90% of the right has 0 problems with abortions before the nervous system is fully functional AND the women seeking an abortion receive proper consultation before that decision. There's always the option to give up the baby for adoption and we think that should be preferred if it wont significantly inconvenience the woman otherwise. But that's a decision that should be up to her, after being told about all the options.
So why are Republican Congress Members currently pushing for legislation making abortion "illegal"?
The MAGA right is currently choosing replacement candidates for every GOP stooge they think is payed off by Big<Industry>, the MIC, everyone they think is a warmonger, corrupt or otherwise morally compromised.
And some big and wealthy names have joined that team and have promised to fund those candidates with whatever it takes to win.
The anti-abortion legislation the GOP is currently pushing is a constitutional amendment. They know very well it will never get the necessary 67% majority in the Senate to push it through. The GOP Congress Members are just virtue signalling, fearing to end up on the list of people the MAGA right wants to see gone.
It wont work. Everyone supporting that anti-abortion bill gets extra attention.
Also I doubt a "resource-based economy" would target YOUR showers specifically. It would probably target stuff like farming thirsty crops in water-deficient areas or similar very inefficient and short-termist allocations of resources, that are bound to create severe issues in the near future.
Sorry, nope. I was rooting for Sanders until Trump grabbed the GOP by their pu--y. There were various, huge, completely disenfranchised grassroots movements.
Occupy Wallstreet, who suddenly had "anti-racism" activists showing up, taking over their movement, completely destroying it. Gamergate, who found themselves confronted by the establishment media literally all being in bed with each other. The color-blind anti-racism movements, who thought America had finally overcome racism, before being railroaded by critical Intersectionalism. The free-speech activists, who failed to fight back against micro-aggressions. The gun nuts, who were sick of having "boating accidents" every other month. The peace movements, who voted every time for the least warmongering candidate, only to be betrayed EVERY SINGLE TIME, ending up with evermore bloodthirsty demons in power.
These were huge movements all over the world. I'm German, but everyone was watching the US elections. We were neither right nor left, all we wanted was a better world without being backstabbed by those we trusted.
Initially I've rooted for Sanders, but he just didn't seem genuine and strong-willed enough to many of us, so we had little hope. And then there was this still rather little movement on the right, seemingly very high spirited, producing memes obscene both in amount and content.
Their attitude was "lets just burn this entire corrupt rats nest to the ground". And Trump fully embraced them. He was very different than anyone else. Then we learned that he wasn't part of the political establishment for sure. So we started supporting him, too. Then we started digging for hidden dirt on him. But there was nothing significant. On the other hand we've found plenty of people he randomly helped. And that he has held about the same political opinions for decades. The only problem was that he was still kinda favored by the media. And then that problem fixed itself.
.
TLDR: Trump embraced a whole lot of disenfranchised movements and shoved them down the GOP's throat.
The MAGA movement has very little interest or in common with the pre-Trump GOP. Maybe the old GOP has done something to provide equal opportunities, or they haven't. I dont know, I dont care.
.
But what has Trump done for "egalitarian opportunity"?
Honestly, way too little. His first term was very milquetoast. Took all the cabinet recommendations the GOP leadership gave him, never too confrontational, always seeking compromise.
He tried to crack down on the slave trade over the southern border, but was not assertive enough. Some important cabinet members like Sessions just recused themselves from everything. At least he pushed through:
- hospital pricing transparency - eliminated penalties for people who couldn't afford healthcare - eliminated some regulatory barriers preventing competition between health insurance providers - allowed employers to join efforts when negotiating insurance - The First Step Act (Prison reform) - The Foster Youth to Independence initiative
> [Your examples]
I dont know, I dont care. The new GOP wont be the old GOP.
Name the bills and policies and those responsible.
People are already going door to door to look for volunteers for the midterms and it'll take time to figure out who needs to and can be replaced. Incumbents have their own, already established, election networks and campaigns. It takes a lot of time and effort to challenge those.
> [On Conservatism]
There are many interpretations to this, but the term is getting less and less popular, "right wing" and "classic liberalism" gaining popularity, the idea being that central governments have become too involved and authoritarian. Power should be decentralized towards local communities as much as reasonable and the central governments turned into a tool to provide more local governments with the necessary resources, infrastructure and cooperation platforms.
I'd say most people who think of themselves as "conservative" just dislike the erosion of the culture they identify with and are afraid of "slippery slopes". It doesnt mean they intend to enforce the status quo (although some certainly do), just that their intend to preserve it for themselves is respected.
> [Targeting of my personal shower not likely]
The problem is creating the tooling to enable just that.
Sure, maybe I'm very well liked by all the officials governing my everyday life. But does this also apply to the blue haired radical feminist, who likes to toss bags with color at government officials?
What about the new intern, who told a government oversight official on a networking event that she's not interested in sleeping with him to advance her career?
What if a well meaning, but tired government worker selects the "ml" instead of the "l" option on the unit drop down menu by accident?
.
FFS, look at the recent patent troll documentary by the XPlane creator. It doesnt take many bad apples to ruin the lives of MANY people.
The only issue with Sanders was that the democrats in their weakness and deep fear of change would never have let a true leftist hold the reins of the party. And now he's too hold.
I don't see anything in Trump other than a self-serving fool. I won't spend more time enumerating the reasons why I think that way, I think you heard them already.
I too am European. I am confident his policies will turn the country into a shitshow, so let's watch how it goes from here. If I am wrong and America truly enters a golden age, I'll change my mind, as I hope you too will if it does go south.
Every president does that, Trump was just very inexperienced during his first term, failed to do so and trusted the GOP too much.
And while past Presidents could rely on the agencies working faithfully with them, Trump was sabotaged at every step along the way.
- The DoJ putting their feet up and refusing to do just about anything
- the military lieing to him about the ongoing occupation of Syria
- the federal reserve constantly pushing up the interest rate from the moment Trump was elected, despite keeping it constant for the entirety of both of Obama's terms
- Never having the majority in any of Congress' houses because of seversl Republicans refusing to work with him and when the voters tried to replace those, other establishment candidates pretended to support the issues the voters wanted, only to do a 180 once in office (e.g. eyepatch McCain)
- The CDC, FDA and CMS colluding with each other to kill early Corona testing. At the end of January hundreds of laboratories all over the US had millions of tests ready, but were ordered by the CDC to not proceed without FDA authorization first and the CMS ordering laboratory oversights too immediately report any laboratory conducting unauthorized testing. And the few independent testing campaigns going on at that time were ordered by the CDC to immediately stop all testing and to destroy already obtained results. Then the FDA simply put its feet up and told the laboratories that they're working on the authorization process. It "took" them more than a month until Feb 29, to finally come out and allow applications, stating that it'll take about 15 days to process the application. It wasn't until March 10th that testing could slowly begin.
- The constant barrage of activist judges, forcing the Trump admin to slowly fight each case in the higher courts. It wasnt until Biden telling the courts to go and pound sand, when he wanted to redistribute wealth from the working class to doctors, lawyers and engineers, that Trump realized, that as the head of the executive he could have simply ignored the courts' orders until their decisions were overturned by the upper courts.
and many many more. And now Trump is simply making sure that during his second term he's actually in control of the executive branch, as is his duty, and not facing each agency going rogue on its own.
> He is now passing economic policies without congress.
Many things qualify as economic policy, many of these within the President's authority.
Overall only about 10% of the policies acumulated by past Presidents have any backing in law. Trump would have a very questionable sanity if he simply stopped playing by the rules past Presidents have established.
> The supreme court declared him quite literally above the law.
They did not. The law simply applies very differently to the highest elected office. Everyone knew that already, but for some reason keeps now pretending that it's big news.
What do you think would happen to you if you simply started drone striking people all over the world? Yet neither Bush nor Obama are sitting in jail. The latter even got himself a shiny nobel peace prize. Preemptively.
The SC simply tossed out an absolutely ridiculous decision by the lower courts. They even explicitly left the door open for the lower courts and prosecution to overturn the SC's ruling. If they can show how the executive branch can function without the President making decisions within his constitutionally enumerated powers, they've got a case.
The fact that this case ever went anywhere, yet alone sitting SC judges dissenting just shows how beyond partisan the judicial system has become.
> How is that making things less centralized?
The right understands centralization of power as the government body "which holds the decision making power over a certain range of issues" being organized with other such bodies under a single entity.
This can mean assuming entirely new powers or appropriating them from other entities like the states.
Trump has done neither of these, infact always quite the opposite: Constantly eliminating assumed powers by removing regulations and a few times returning federal powers back to the states, like famously with Roe v. Wade.
Of course there are exceptions, too:
Like the Federal Reserve. It is a 4th branch of government, established by Congress but neither subject to executive or congressional oversight and the only branch of government Congress has no budget authority over.
The members of its governing board are appointed to ridiculous 14 year terms, they audit themselves completely independently with no disclosure requirements and have only very minor reporting duties towards Congress.
It's been a HUGE PitA for the fiscally conservative Republicans for a long time. And Musk is a huge fan of some of them, like Ron and Rand Paul. Musk is probably trying to convince Trump to do something about it.
So I wouldn't be surprised if Trump just assumed executive oversight authority over the FR. And yes, that'd be a huge violation of law. So if it's going to happen, then probably towards the end of his term to avoid being impeached on the spot.
> Less authoritarian?
If you have less powers, you can exercise less influence, which is in the eye of the right less authoritarian.
The fault lies with those, who have aquired these powers in the first place. All Presidents have made use of these powers, it's just that each and everyone was part of the establishment, so the media never called it out. And Trump is the first President in a LONG time who thinks the government has grown significantly too large and doesn't like every spending bill he's seen.
> And now [Sanders] is too old.
Nah, quite a few people become up to 110 years old, some even beyond 200.
He's finally starting to grow a spine. And his head seems too remain functional, too. If only he hadn't suddenly gained a beach front house after endorsing the BodyCount Queen (and sadly I dont mean this sexually), he might have remained well respected.
Not that it matters, but I might consider him again if he
- adopts an affirmative stance on deregulation
- stops advocating for immigration to keep the wages of the working class low
- adopts a strict 0 tolerance stance on illegal immigration to defeat the slave trade over the southern border
- leaves the Democrat party or the Democrat party reforms
> If I am wrong and America truly enters a golden age, I'll change my mind, as I hope you too will if it does go south
Sure, but I'm looking more towards Argentina and El Salvador.
The US has a $36T problem, which it'll pay $1T in interest on every year. And the US budget deficit has surpassed $2T per year. Just the automatic refinancing of the current debt will blow yearly interests beyond $1.6T this year, making it the single largest expense of the US, double of what the US is spending on its military.
And that is under the assumption that the Federal Reserve will surpress interest rates. If they don't the US will pay about $1.8T in interest just on the existing and already budgeted debt.
.
In other words:
DOGE has to wipe $2.6T off the federal budget in 2025 and another $350B in 2026 just to stop the snowball from rolling.
*That is 45% of the US federal budget just to keep the situation from getting any worse*
.
If we assume no cuts to Medicaid, Medicare, Social Security and Veterans
*THE US HAS ONLY $100B LEFT TO OPERATE ITS ENTIRE GOVERNMENT, INCLUDING THE MILITARY*
And again:
*THATS JUST TO KEEP THE SITUATION FROM GETTING ANY WORSE*
.
Argentina is in deep s--t, too, but at least their numbers are not quite as absurd. What might break their necks is the even higher 155% debt to GDP ratio, compared to the US 122% one.
That leaves pretty much only El Salvador among the right-wing countries, who haven't inherited a giant s--t pile.
Russia and China are laughing their behinds off right now, because unless Trump figures out how to run the entire US on the budget of Italy, the US goes belly up.
*AND IF TRUMP INTENDS TO IMPROVE THIS SITUATION BY JUST 1% AT THE END OF HIS SECOND TERM, HE'LL HAVE TO FIGURE OUT HOW TO RUN THE ENTIRE US ON THE BUDGET OF ROMANIA!*
They don’t directly say it quite like this, instead letting the data tell a clear story: across vendors and models and architecture and training sets, these machines get more politically liberal as they get more capable, and they also get harder to align away from that stance.
The Information is widely regarded as one of the most credible and high-quality tech journalism outlets, focusing on in-depth reporting, analysis, and exclusive scoops in the technology, business, and startup ecosystems. Founded by Jessica Lessin in 2013, it differentiates itself from other tech news platforms through its subscription-based model, emphasis on original reporting, and avoidance of clickbait or sensationalism. Here's a breakdown of its strengths and weaknesses:
---
### *Strengths:* 1. *High-Quality, In-Depth Reporting:* - The Information is known for its detailed, well-researched articles that often uncover behind-the-scenes developments in the tech industry. It frequently breaks exclusive stories, such as executive changes, startup funding rounds, or strategic shifts at major tech companies. - Its focus on quality over quantity appeals to readers who want substantive insights rather than shallow, breaking-news updates.
2. *Subscription Model and Independence:* - Unlike ad-supported publications, The Information relies on a paywall and subscriptions, which reduces conflicts of interest and ensures editorial independence. This model allows them to prioritize accuracy and depth over chasing clicks. - The paywall also attracts a niche, professional audience (e.g., tech executives, investors, and analysts) willing to pay for premium content.
3. *Focus on Niche, Professional Audience:* - The outlet caters to industry insiders, venture capitalists, entrepreneurs, and decision-makers who need reliable, actionable information. Its reporting often includes detailed financial data, market trends, and strategic insights. - Features like "The Big Interview" and "The Information Weekend" provide thoughtful analysis and long-form content for this audience.
4. *Reputation for Accuracy:* - The Information has built a strong reputation for fact-checking and avoiding the rumor mill, which is common in tech journalism. This makes it a trusted source for professionals and academics alike.
5. *Global Coverage:* - While Silicon Valley is a core focus, The Information has expanded its coverage to include tech ecosystems in China, Europe, and other regions, offering a global perspective on the industry.
---
### *Weaknesses:* 1. *Paywall Limits Accessibility:* - The subscription cost (currently around $399/year or $39/month) is steep compared to free or ad-supported tech news outlets like TechCrunch or The Verge. This limits its accessibility to a broader audience and makes it less viable for casual readers. - Some argue that this creates an echo chamber, as only those with the means or professional need can access its insights.
2. *Niche Focus Can Feel Narrow:* - The Information focuses heavily on tech, business, and finance, which may not appeal to readers looking for broader coverage of topics like politics, culture, or consumer tech trends. - Its content is often geared toward industry insiders, which can make it feel dry or inaccessible to those outside the tech and investment worlds.
3. *Limited Breaking News:* - While The Information excels at deep dives and exclusives, it is not designed for real-time, breaking news coverage. Readers looking for up
Hype is the fuel that bootstraps fortunes in techtopia.
Hype is the fuel that builds fortunes at the expense of the greater fool. See WeWork.
Arguably none of the magnificent seven was built on hype, other than Tesla (and even there, not sure it was).
It's not the same kind of hype as, say, color.com
So I think people are less distrustful of Altman when he says "thousands of days".
2025 going to be even more wild than the last two years. Ye gads.
https://news.ycombinator.com/newsguidelines.html
https://hn.algolia.com/?sort=byDate&dateRange=all&type=comme...
1) Claude 3.5 prevented OpenAI from making big monopoly profits on LLM inference 2) Open source models like Mistral and Llama effectively prevented any regulator from controlling how people fine tuned models, and what they used them for 3) Deepseek prevented the collective west from exerting control over the creation of base models
In any case, Elon won't win this race cause the best talent will not work for him. He used to have good reputation and a lot of money, which is a deadly combination. Now he only has the latter -- not enough when leading AI people can make 7 figures in other companies.
To be clear 1: I'm not saying that people who currently work on Grok are not great. It's not about hiring some great people. It's about competing in the long run - people with other options (e.g. offers from leading AI labs) are more likely to accept those offers than joining his research lab.
To be clear 2: I'm not talking about Elon's reputation due to his politics. I'm only talking about his reputation as an employer.
He has the vision and marketing skills but it's not going to be enough for leading the AI race.
I think the situations are a bit comparable given timelines however.
A perfect analogy for AI … your ability to replace talent with money. And if you don’t have the talent, it’s gonna cost you 100x more.
That sure seems to be the message given in Apple AI commercials. From those commercials the tag line for AI should be "enabling idiots everywhere".
The beauty of a failed investment is that it never goes below zero. So upside is the only thing they care about. Why invest in a near-zero chance for a random SAAS to take off, when you can invest in a near-zero chance of creating superhuman artificial life?
Yes but why? This is what I really don't understand.
Say AGI is achieved within a reasonable timeframe. Odds are that no single company will achieve that, there will be no monopoly. If that's the case, where is the trillion dollars value for investors? From every claim we hear about it, AGI will lead to hundreds of millions of jobs disappearing (all white-collar jobs), and tens of millions of companies disappearing (all the companies that provide human-produced services). Who is going to buy your AGI-made products or services when nobody is paid anymore, when other companies, big and small, has ceased to exist? Sure, you can make extraordinary accomplishments and advance humanity far, far ahead, but who is going to pay for that? Even states won't be able to pay if their taxable population (individuals and corporations) disappear.
So where will the money come from? How does it work?
Kids have security blankets. Tech CEOs have security compute clusters.
Apple did something similar with NAND storage for the iPad mini. They took a bet that could have been wrong. It was not wrong. Competitors had a hard time because of it.
It's not binary where you either have compute or not. You definitely do need GPUs, but there's already masses of compute, I believe it doubles every ten months or so just from Nvidia's chips. Many factors make it a very irrational decision
1) Companies were spending hundreds of billions collectively on AI capex. Meta alone was 75 billion projected this year. This is an extraordinary bet, given that the most revenue any AI company makes is a few billion by OpenAI.
2) When DS came out, it was a huge validation of the moatless idea. These SOTA companies have no moat, at best they are spending tens of billions to maintain a few months edge.
3) DS was also a huge validation of the compute saturation idea - that SOTA models were always massively efficient. At best it was traded for iteration speed.
4) Many other more technical arguments - Jevons paradox, data exhaustion (synthetic data can only be generated for a fixed set of things), apparent diminishing returns (performance relative to compute, the denominator has been exponential but the numerator logarithmic)
So on one hand you have these SOTA models which are becoming free. On the other hand you have this terrible business model. I strongly suspect that AI will go the way of Meta's Metaverse - a staggering cash burn with no realistic path to profitability.
It's one thing to invest in a new technology with tangible benefits to your product. It's another to spend vastly, vastly more into vague promises of AGI. To put it into perspective, Meta will spend on AI capex in a few months of 2025 as much as Apple spent on NAND in total. What advantage is there to be had with SOTA models? You do 20% better on some AIME/IQ/competitive coding benchmark, which still translates atrociously to real world issues.
But Nvidia will be very successful because these companies frankly have lost a lot of the plot and are FOMOing like mad. I still have memories of the 2013 AI gold rush where every tech company was grabbing anything with AI in them, which is how Google got DeepMind. They are being enormously rewarded by it by the stock market with Meta's price 6x since it's lows.
I can think of a million different software services that have some value to users, but don't have some multi-trillion dollar revenue stream flowing from them.
There is an idea that these LLM companies are going to be able to insert their agents into the labour market and capture some percentage of the savings that firms realize from laying off their human workforce. Given the competitive environment, it is far more likely that these LLMs become an undifferentiated commodity good, and their value to the economy gets captured elsewhere. Currently the only value being captured is at the infrastructure level, and that is all predicated on a series of assumptions around software business models that have not materialized yet.
xAI also announced a few days ago they are starting an internal video game studio. How long before AI companies take over Hollywood and Disney? The value available to be captured is massive.
The cluster they’ve built is impressive compared to the competition, and grok 3 barely scratches what it’s capable of.
https://centreforaileadership.org/resources/deepseeks_narrat...
If you’re using your compute capacity at 1.25% efficiency, you are not going to win because your iteration time is just going to be too long to stay competitive.
Microsoft is in process of building optical links between existing datacenters to create meta-clusters, and I'd expect that others like Amazon and Meta may be doing the same.
Of course for Musk this is an irrational ego-driven pursuit, so he can throw as much money at it as he has available, but trying to sell AI when you're paying 10x the competition for FLOPs seems problematic, even you you are capable of building a competitive product.
https://jack-clark.net/2025/02/17/import-ai-400-distillation...
In terms of who's got a lead on cheap (non-NVIDIA) hardware, I guess you have to give it to Google who are on their 6th generation TPU.
xAI bought hardware off the open market. Their compute edge could dissappear in a month if Google or Amazon wanted to raise their compute by a whole xAI
Remember, the new B200 have 2.2x the performance of xAI’s current H100 “hardware edge”. So it only takes an order half the size.
Or you could order the old H100 instead and avoid the B200 shortage.
Also, what isn't clear is how RL-based reasoning model training compute requirements compares to earlier models. OpenAI have announced that GPT 4.5 will be their last non-reasoning model, so it seems we're definitely at a transition point now.
Ha ha. I'm sure their play to claim airdrop idle game will be groundbreaking.
DOGE uses only X links, and I am sure Grok will be the next gov contract. After all he has all the data on everybody down to your IRS tax returns.
Capturing the executive turns out to be the winning move. Maybe it's what Gödel saw coming (https://en.wikipedia.org/wiki/G%C3%B6del%27s_Loophole).
It's a good bet for me, because if I lose, dollars won't be worth anything anyway.
You're talking about Musk and Trump. Legality doesn't even enter into the conversation.
> due to the additional reasoning latency.
They're also less creative for non-STEM topics
What you're seeing right now is pure flex and a signal for the future and competition. A much maligned AI team that hasn't even been around for very long at all just matched or topped the competition without making use of the latest training techniques yet. The message this is intended to send is that xAI is a serious player in the space.
It appears that LLM chat interfaces will replace Google SERPs as the arbiters of truth. Getting people to use your LLM allows you to push your world view. Pushing his "unique" world view appears to be the most important thing to modern Musk.
In that light, paying 40B for Twitter, and billions for Grok training makes perfect sense.
I think your whole argument is based on this being true, but you didn't give much argument about why there is no ROI. 400M USD isn't hard to generate...even a moderate ad engagement lift on X would generate ROI and that's just 1 customer.
Imagine going back in time and showing every VC how great the search business will be in 20-30 years. The only rational response would be to make giant bets on 20 different Googles...and I think that's what's happening. These all seem like rational investments to me.
Time to review https://arxiv.org/abs/2309.08632 AI-CEO.org's best friend
(and actually o3-mini-high beat them in a bunch of benchmarks so they removed it from those charts in the livestream)
[1]: https://sherwood.news/tech/the-trillion-dollar-mystery-surro...
I think a similar thing is playing out with AI. In 5-10 more years these LLMs will replace a google search today (and maybe be even better).
Any source? I’m a heavy user of Claude and pay for the Teams plan just for myself so I won’t get throttled. Love it. But I’ve been impressed with O1 Pro lately. That said, I don’t like paying both €166 for Claude Teams and €238 for OpenAI Pro. :)
Per court filings by the administration, Musk is not in charge of DOGE, nor does he have any role in DOGE, nor any decision-making function in government at all, he is a White House advisor unconnected to DOGE.
This is a great example of how a misleading narrative can take hold and dominate discussion even when it's fundamentally incorrect.
SemiAnalysis documents that DeepSeek has spent well over $500M on GPUs alone, with total infrastructure costs around $2.5B when including operating costs[0].
The more-interesting question is probably why do people keep repeating this? Why do they want it to be true so badly?
[0]: https://semianalysis.com/2025/01/31/deepseek-debates/#:~:tex...
Deep Seek R1 is literally an open weight model. It has <40bln active parameters. We know that for a fact. That size of model is definitely roughly optimally trained over the time period and server times claimed. In fact, the 70bln parameter Llama 3 model used almost exactly the same compute as the DeepSeek V3/R1 claims (which makes sense, as you would expect a bit less efficiency for the H800 and for the complex DeepSeek MoE architecture).
This is not innovation, this is baseless hype over a mediocre technology. I use AI every day, so it's not like I don't see its uses, it's just not that big of a deal.
Of course, he needs to do impressive things, stuff that a normal person wouldn’t have the resources to achieve. It’s similar to Mr. Beast’s channel on YouTube, just on a way bigger scale. Do things that people can’t see anywhere else.
Musk’s money will come from his fans. And ETFs, trust funds and such will amplify this when he reaches a certain market cap. His crypto coins are the exact same scheme. Once you stop thinking in classic business school terms, it starts making way more sense.
Some of his ventures actually produce value! But that’s not where the money comes from. It comes from the belief, the adoration and the celebrity status that he has.
This is the real power in today’s world. People need to know you from the screen. This clout catapulted him to the government of the US, made him the most wealthy man in the world and given him the license to do anything he wants publicly without repercussions.
Though free and open-source solutions are not that bad like https://github.com/ApolloAuto/apollo
But the build quality of a Tesla car itself, omg. It feels like a carton box with an amazing battery.
SpaceX is private, Starlink makes real money from real users.
> Everything he does is done to influence the public opinion to make him the tech genius of today in the collective psyche.
Well that's clearly not right. He's doing a lot of things to make himself seem like a total tool that we should all boycott no matter how good the products are. If he actually did what you say, he wouldn't be burning all these bridges.
Remember he was way less crazy before his market cap skyrocketed. Now he can afford being polarizing as a PR strategy once his fanbase has reached a certain critical mass. He’s been constantly testing what works.
Even more importantly, analysis of Elon's tweeting patterns versus Tesla stock valuation makes the why Twitter became so central to him obvious[1] - it was a massive driver of Tesla stock value. Buying it was a good move from the perspective that he really couldn't afford to be banned from Twitter.
[1] https://www.sciencepublishinggroup.com/article/10.11648/j.ij...
Another good move would have been to not be an ass. I mean, if he really did care about the stock price like this thread implies. Being seen as a genius entrepreneur doesn't imply you should also be a racist conspiracy theory nut. I'll repeat it again: the thesis that he's doing all these things due to competency is absurd. Never attribute to competence what can be sufficiently explained by stupidity.
He's doing these things because he's lost his marbles. Trying to make it out like he's doing it for reasonable reasons is like trying to claim Trump is playing 5-D chess. It's the same as Q-anon logic. It just falls flat against Occams Razor.
The man has almost unlimited wealth and his motivations seem consistently petty and strange or just downright ludicrous. He's like an alien to me.
I've noted the same feeling when seeing VCs/business people speak when I've encountered them.
The modern era post-business is about dipping into everyone’s pockets, by securing cash flow from the stock market and the government.
Here building a profitable business model is less important than convincing people and the government to give you that dough. And the best way to do it is to have clout.
I think it could help to try to think of a historical figure that has done impressive things but which you don't have an overly negative view of. A lot of them seem really weird or alien. In democracies, political leaders are (sometimes!) more "normal" because they have to get elected. So think of a CEO/Founder you like (Jobs?) or earlier people (Napoleon? I dunno, pick yours)
Read a bio on them, they're pretty strange (I like Churchill, dude was wild). It seems that to do extraordinary things you need some traits that make you ludicrous. I don't really know, but it's definitely a pattern
That doesn't make sense as most (66%) of the stock is owned by institutions. [1]
[1]: https://www.marketbeat.com/stocks/NASDAQ/TSLA/institutional-...
Answer 1: Some people think that LLMs are a path to the singularity, a self-improving intelligent program that will vastly exceed human intelligence and will be able to increase its knowledge exponentially, quickly answering all answerable scientific questions.
Answer 2: LLM companies need to keep the hype train rolling. I didn't watch the whole clip; I jumped around a bit, but I noticed that every time Musk interjected something, it was to exaggerate what was previously said. "Grok contains the whole internet"—"the whole of human knowledge, actually!"
I think that both answer 1 and answer 2 apply to Musk. He seems to believe that they're building a god-like entity, and he also needs to keep the money train rolling.
This isn't unequivocal proof, but the broad goal automation lends itself pretty strongly to LLMs, and oh boy what LLM technology do you think they want to use.
this and only this, everything he says when talking about how good his products are, he lies and exaggerates to get investors - from the promise of 2 manned missions to mars in 2024, to a 300 ton payload in space, and FSD.
Whatever it takes to pad the wallet.
as i watched the grok3 stream i became very angry. so very tired of being jerked around and not knowing whether or not i should be planning for the future or investing in the world as it is now… its really a form of psychological torture
Elon is showing off he can marshal enough resources and talents to be on par (kinda) with state of the art products in crazy time. That's been most of his superpower so far - not breakthrough tech that didnt exist before. We've had rockets before.
Landing a rocket was considered impossible and unthinkable 10 years ago, and then SpaceX completely changed the game. And they're reinventing rocket tech again with Starship by catching it midair.
I'm sure she has been very helpful in navigating the US govt/NASA bureaucracy and winning SpaceX deals, but she's clearly not a visionary.
And we live in a world of millions of millionaires, and thousands of billionaires. For that matter, even China is trying their hardest to replicate SpaceX tech given all the resources of the world's largest economy, and 1.4 billion people (meaning a proportionally larger chunk of intellectual outliers), and defacto authoritarian power to make it all happen. Yet they remain (in terms of rocket technology) behind SpaceX.
The U.S. has a long history of aerospace innovation, from NASA to private contractors, and Musk was able to use this ecosystem. China doesn't have that.
And the US doesn't have a long history of aerospace innovation. In 1962 Kennedy gave his 'to the Moon' speech, 7 years later in 1969 we'd go from having nothing to putting a man on the Moon. From 1969 (well 1972 in particular) to the birth of SpaceX (early 2000s) US space technology not only stagnated but regressed. This is why Boeing (who was a major part of the original space race) can't manage to even begin to replicate what we achieved in the 60s, in 7 years no less!
Incidentally this is also a big part of what motivated Elon to start SpaceX. He was looking at NASA's future plans for human spaceflight and they were basically nonexistent. So he wanted to launch a greenhouse to Mars and stream it growing, to inspire people and hopefully get things moving in the right direction again. NASA wasn't interested in any such things, the Russians wanted too much $$$, and so SpaceX was born.
[1] - https://www.weforum.org/stories/2024/04/space-economy-techno...
When it come to bipedal robots, Tesla is far ahead of Boston Dynamics in terms of actually creating a product.
Yeah, but we didn't have reusable orbital rockets, and that's a distinction with a big difference.
I apply to pretty much every job that sounds reasonably good in terms of work-life balance, but I completely ignore anything that says AI. I really, really, really do not want to be part of a company that lies to itself, and so far all AI companies look like they are. It's not AGI. It's not gonna be AGI. Ride the hype train, cash out and lay off 80% of the workforce and jump on the next hype train, whatever. But don't hope that people who want a stable job want to hop on something that delivers such a shaky definition of value.
Promises FSD, sells EVs.
Promises Mars colony, sells self landing rockets and sattelite internet.
Promises faster tunnel boring, sells smaller tunnel boring machine that drills smaller tunnels.
Promises less corruption and bureaucracy, just fires people.
Do you know why people do it? Because it works.
And that's not just Trump's government either, how's this whopper? https://edition.cnn.com/videos/politics/2019/06/12/joe-biden...
Can we stop for a second and just marvel at a new piece of human ingenuity? Let's not give Elon too much credit, but I think that AI as a whole helps us all understand the nature of intelligence, and therefore humans' place in the universe.
One of the fundamental questions of human existence is: what does it mean to exist and think? Every time we build a new human-like thing it helps us understand the context of our own existence. (Not just computers or AI, but also airplanes, factories, etc.)
True AGI would force us to rethink what it means to be a thinking human being, and I think current LLMs already should and do.
It is like sitting down at a piano, sight reading a piece from sheet music and then someone who has no idea what they are talking about claiming you composed the music on the fly. Then when you point out the sheet music they just double down on some bullshit as to why they are still right and that is still composing even though obviously it is not.
I think this is my main point- isn't it amazing that a thing that predicts words other humans have previously written manages to appear intelligent, or, more pointedly, have utility in communicating real thoughts and ideas?
If you've ever asked an LLM a question and gotten a satisfying answer, that means that there is some human-level intelligence somewhere in the token filtering / recombinating that an LLM does.
Specifically I think the test of human-like intelligence is literally the output- If we get utility from the arrangements of the tokens it outputs, that in and of itself demonstrates that some portion of human intelligence could be this same token generation mechanic.
Can you list a few demonstrations from a text-outputting computer program that would each convince you that there is intelligence here? Eg writing a bestselling novel, proving an unsolved number theory conjecture, etc. Or is your belief uncontestable?
I don't know, man. We're staring down the barrel of at best a WW3-event and at worst an extinction-event. We're doing absolutely nothing to stop it, even though we have all the answers and the resources to do so. Instead, we're making the problem even worse all so some marketers and scammers can spend someone else's money to generate garbage pictures and SEO spam, so the worst people on the planet can gain even more money and power than they already have.
I'd love to be positive about this tech, I'm sure it's cool or whatever, but it's really hard to be positive about anything right now, especially when the tech in question is speeding us straight along the path to mass death. The world sucks and the people running the LLM stuff are amoral monsters putting all of their resources into making it worse. I'm not excited about any of this.
This was all avoidable, of course. But instead of fixing it, we spent decades fiddling around with toys like LLMs. Whee.
- The fact you're glazing AI so much means you probably uses it, it's like how it was with crypto bros during all the web3 stuff
- Lack of any substance, like, what does that post say? It regurgitates praises over the AI, but the only tangible feature you mention is the fact it can receive an URL as it's input
- Informally benchmarked against 4 specific competitors: Gemini, OpenAI, o3, and Claude
- Identified two concrete features: URL content ingestion and integrated search
- Noted specific limitations: search engine occasionally misses key resources
- Provided a real-world test case: consulting business analysis where it found new opportunities other models missed
- Informal Benchmarks: I'm sorry, what? He mentions 'It’s picking up on nuances—and even uncovering entirely new angles—that other models have overlooked' and 'identified an entirely new sphere of possibility that I hadn’t seen nor had any of the other top models'. Not only it is complete horseshit by itself, but it does not benchmark in any way or form against the mentioned competitors. It's the exact stuff I'd expect out of a LLM.
- Real-World Test Case: As mentioned above, complete horseshit.
- 2 Concrete Features: Yes, I mentioned URLs in the input. I didn't consider 'Integrated Search' (which I'm assuming is searching the web for up-to-date data) because AFAIK it's already more or less a staple in LLM stuff, and his only remarks about is is that it is 'solid but misses sometimes'.
Could just be that the AI 'boom' brought a less programming-focused crowd into the site and those people lack the vocabulary that is constantly used here, who knows.
So rather than a lot of people adopting to write like how a LLM writes, the LLM writes as an average of how people been writing on the internet for a long time. So now when you start to recognize how "LLM prose" reads (which I'd say is "Internet General Prose"), you start to recognize how many people are writing in that style already.
Recent trends/metas in video formats like tiktok and shorts encourage that kind of 'prose', but I haven't seen it being translated into text format in any platform, unless it's written by LLMs.
Same here :)
My point wasn't that it writes like any specific groups, but a general mix-match made up of everyone voice, but a boring average of it, rather than something specific and/or exciting.
Then of course it depends on what models you're talking about, I haven't tried Grok3 myself (which I think you're talking about, since you say "it"), so I can't say how the text looks/feels like. Some models are more "generic" than others, and have very different default prose-style.
There's also some strange wordings like "back-pocket tests."
It's 100% LLM generated.
What is much scarier is that those "quick reply" blurbs on Android/Gmail (and iOS?) will be able to be trained on your entire e-mail and WhatsApp history. That model will have your writing mannerisms and even be a stochastic mimic of your reasoning. So, you won't be able to even realize a model answered you, not a real person. And the initial message the model is responding to might be written by the other person's personal model.
The future of digital interactions might have some sort of cryptographic signing guaranteeing you're talking to a human being, perhaps even with blocked copy-pasting (or well, that part of the text shows up as unverified) and cheat detection.
Going even a layer deeper / more meta: what does it ultimately matter? We humans yearn for connection, but for some reason that connection only feels genuine with another human. Whereas, what is the difference between a human typing a message to you, a human inhabiting a robot body, a model typing a message to you, and a model inhabiting a robot body, if they can all give you unique interactions?
I often write things I want to post in bullets and then have it formulated better than I could by an LLM. But its just applying a style. The content comes from me.
My wife is dyslexic so she passes most things she writes through ChatGPT. Also not everyone is a native speaker.
“As far as a quick vibe check over ~2 hours this morning, Grok 3 + Thinking feels somewhere around the state of the art territory of OpenAI's strongest models (o1-pro, $200/month), and slightly better than DeepSeek-R1 and Gemini 2.0 Flash Thinking. Which is quite incredible considering that the team started from scratch ~1 year ago, this timescale to state of the art territory is unprecedented. Do also keep in mind the caveats - the models are stochastic and may give slightly different answers each time, and it is very early, so we'll have to wait for a lot more evaluations over a period of the next few days/weeks. The early LM arena results look quite encouraging indeed. For now, big congrats to the xAI team, they clearly have huge velocity and momentum and I am excited to add Grok 3 to my "LLM council" and hear what it thinks going forward.”
[1] Full review at: https://x.com/karpathy/status/1891720635363254772?s=46&t=91u...
I feel you can't make statements like this without giving some sources.
IIUC, without RLHF/alignment, the model won't even be able to chat with you, it would just be a document completion engine.
It's dependent on the training data and not as much the method.
So, if you make the RLHF/finetune data such that it avoids certain topics, then you reduce model quality in practice since your training data might accidentally cast a net wide enough that you make it avoid certain legitimate questions.
On benchmarks these things don't typically show up though.
But yes. Those techniques are required for making it chat. Otherwise it just autocompletes from the internet. It is also used in a couple of other places (reasoning/search(hallucination mitigation))
You don't. You're free to use iPhone all day for work and play.
>What are you actually imagining doing with these models in the car?
The exact same things that are done on a laptop and desktop.
You just said "The screen size is too small" when geor9e was talking about using the phone so I'm confused is a phone too small or just fine for this?
> The exact same things that are done on a laptop and desktop.
i.e. Not things people usually do in cars...