> DRAM is priced based on supply and demand, like every other market.
Please don't explain it away like that - you are referring to the theoretical "ideal" market where a bunch of small companies compete with low margins to the benefit of the wider customer base. This is not what is happening. We have a couple of intrinsically worthless, LLM-whale companies, working literally to swallow and entshittify literally everything in their weird transhumanist/accelerationist/weirdo way. To add to the insult, the whole creation of artificial scarcity is almost a political construct, paid for with "monopoly-the-game-money" that these companies DO NOT EARN but instead BORROW based on vague and dishonest promises of achieving a "Country of PhDs in a datacenter"/"Pocket PhDs"/"AGI by 2025" (oops, now apparently by 2028 according to the OpenAI CEO). In their weird vision, as humans we should be merely cattle to be managed, not independent spirits with interest and aspirations. That ghoul Karpathy speaks about "ghost in the machine", overlooking the magnificence of the already existing "ghost in the machine" in the form of human beings. We should not have to swallow the increasingly crappier future these folks are insisting on pushing on all of us.
Right, because as seen over the last several years, the Big Tech CEOs should totally be trusted on their promises, especially if it is related to how our sensitive personal data is stored and processed. This goes even wtihout knowing who is one of the better known "personas" investing in Persona.
It is a rather attractive view, and I used to hold it too. However, seeing as Alphabet recently issued 100-year bonds to finance the AI CapEx bloat, means they are not that far off from the rest of the AI "YOLO"s currently jumping off the cliff ...
Well Bill Gates declared it 20 years ago : "Content is king", so here we are. "Content" everywhere, and not just the Internet - go to anny tourist hotspot, there is a ratio of 5:1 of idiot influencers all stretching their faces in front of the same landmarks. Your utility company now produces "content" for some reason. Every lazy Tom, Dick and Joe produce "content" instead of doing their trade. Apps that summarise your books so you can "read" them on your commute. To the ghouls who are driving the "content" economy, everything is "content". They don't really understand either long form or novels, or music or anything creative. To them it's all just content, to be bought, sold and consumed by the pound.
> This excuse is hollow to me. In an organization of this size, it takes multiple people screwing up for a failure to reach the public, or at least it should.
Completely with you on this, plus I would add following thoughts:
I don't think the size of the company should automatically be a proxy measure for a certain level of quality. Surely you can have slobs prevailing in a company of any size.
However - this kind of mistake should not be happening in a valuable company. Microsoft is currently still priced as a very valuable company, even with the significant corrections post Satyas crazy CapEx commitments from 2 weeks ago.
However it seems recently the mistakes, errors and "vendors without guidelines" pile up a bit too much for a supposedly 3-4T USD worth company, culminating in this weird random but very educational case. If anything, it's indicator that Microsoft may not really be as valuable as it is currently still perceived.
This is hilarious actually. I am starting to lean into "AI-dangerous" camp, but not because the chatbot will ever become sentient. Its precisely because of increasingly widespread adoption of un-reliable tools by the incompetent but self-confident Office Worker (R).
It's funny how they and OpenAI keep releasing these "minor" versions as if to imply their product was very stable and reliable at a major version and now they are just working through the backlog of smaller bugs and quirks, whereas - the tool is still fundamentally prone to the same class of errors it was three "major" versions ago. I guess that's what you get for not having a programmer at the helm (to borrow from Spolsky). Guys you are not releasing a 4.6 or a 5.3 anything - it's more likely you are still beta testing towards the 1.0.
> So saying "it predicts the next word" is a nothing-burger. That a program calculates its output one token at a time tells you nothing about its behavior.
Well it does - it tells me it is utterly un-reliable, because it does not understand anything. It just merely goes on, shitting out a nice pile of tokens that placed one after another kind of look like coherent sentences but make no sense, like "you should absolutely go on foot to the car wash". A completely logical culmination of Bill Gates' idiotic "Content is King" proclamation of 20 years ago.
Well, you do understand the "penalising" or as the ML scientific community likes to call it - "adjusting the weights downwards" - is part of setting up the evaluation functions, for gasp - calculating the next most likely tokens, or to be more precise, tokens with the highest possible probability? You are effectively proving my point, perhaps in a bit hand-wavy fashion, that nevertheless still can be translated into the technical language.
You know that when A. Karpathy released NanoLLM (or however it was called), he said it was mainly coded by hand as the LLMs were not helpful because "the training dataset was way off". So yeah, your argumentation actually "reinforces" my point.
> They have an inability to have a strong "opinion" probably
What opinion? It's evaluation function simply returned the word "Most" as being the most likely first word in similar sentences it was trained on. It's a perfect example showing how dangerous this tech could be in a scenario where the prompter is less competent in the domain they are looking an answer for. Let's not do the work of filling in the gaps for the snake oil salesmen of the "AI" industry by trying to explain its inherent weaknesses.
> This thread is extremely negative - if you can't see the value in this, I don't know what to tell you.
This sounds a lot like that line from Microsoft's AI CEO "not understanding the negativity towards AI". And Satya instructing us to not use the term "slop" any more.
Yes we don't see value in taking a git primitive like "commit" and renaming it to "checkpoint". I wonder whether the branches going to be renamed to something like "parallel history" :)
At this point it seems he is not merely excited with the first results (we all were fooled by this tech in the beginning, after all), but actively disseminating falsehoods. After disseminating such claims about AGI repeatedly and spreading nonsense like using ChatGPT to get answers about how to raise a baby day-to-day, which I think no one believes he does - its a statement directed at influencing the behaviour of the so-called "normies", both him and his immediate team should be held personally criminally responsible for every instance of negative impact on human lives that these tools have caused.
For the huge amounts of capital already burnt, and another 1T in CapEX RPO being announced last few weeks and months, isnt that too many of "potentially" and "ultimately" unspecific qualifiers you are throwing around here? Reminds a lot of Sam Altmans classic lines of unspecific statements like "Codex is so good" or "I can only imagine how good it will get" by the end of 202x (insert year of the decade according to your own preference). After 10+ years of OpenAI and 4+ years of ChatGPT, why is the potential not materialising ?
It seems at this point, everyone and their mother, i.e. "We", are building the "tools" for which "we" mostly hope that the VC money will materialise. Use-cases are not important - if OpenAI can essentially work with Monopolly money, whey can´t "we" do it too?
Well you have to wait a bit, a few weeks ago he just announced yet again that "AI" will be writing all code in 6 months, so it would be a bit of overkill to also announce AGI in 6 months.
I wonder how much of the violation of ethical, and often even legal constraints in the business world today one could tie not only to the KPI pressure but also to the the awful "better to ask for forgiveness than permission" mentality that is reinforced by many "leadership" books written up by burnt out mid-level veterans of Mideast wars, trying to make sense of their "careers" and pushing out their "learnings" on to us. The irony being, we accept being tought about leadership, crisis management etc by people who during their "careers" in the military were in effect being "kept", by being provided housing, clothing and free meals.
> Just tested the new Opus 4.6 (1M context) on a fun needle-in-a-haystack challenge: finding every spell in all Harry Potter books.
Clearly a very useful, grounded and helpful everyday use case of LLMs. I guess in the absence of real-world use cases, we'll have to do AI boosting with such "impressive" feats.
Btw - a well crafted regex could have achieved the same (pointless) result with ~0.0000005% of resources the LLM machine used.
You know things are running bad when they've lost even the likes of WSJ. And yet Satya and his ilk (across the LLM space) keep pushing it. It does not matter that this technologically on its own can never reach what they kept promising. Is the pride and ego of a couple of CEOs so important that we'll all just watch them bringing the whole market down ?
Why does this "study" feel like those "studies" from the 50s about the benefits of smoking, which were later discovered to have been (surprise!) financed by the nicotin industry ?
"For example, to create a data visualization I won’t seek any SaaS. I’ll just code one myself using many of the popular vibe coding tools (my team actually did that and it’s vastly more flexible than what we’d get off-the-shelf)."
That maybe doable in your 10-people startup, Namanyay. Try doing it in a larger organisation with layers upon layers of firewalls, databases, authentication systems and not the least importantly - management. Not to mention the vastly different audience, both in size and interest. Your own experience is not the experience of everyone else.