OpenAI: Start using ChatGPT instantly | Hacker News Reader

OpenAI: Start using ChatGPT instantly

openai.com

135 pointsJosely154 comments

> We may use what you provide to ChatGPT to improve our models for everyone. If you’d like, you can turn this off through your Settings - whether you create an account or not.

A quick moan about this:

As a long-time paying subscriber of ChatGPT (and their API), I am extremely frustrated that the "Chat history & training" toggle still bundles together those two unrelated concepts. In order for me to not have my data used, I have to cripple the product (the one I'm paying for) for myself.

It's great that they're making the product available to more people without an account but I really wish they would remove this dark pattern for those of us who are actually paying them.

You can opt out of training while keeping history turned on in the privacy center at https://privacy.openai.com/policies.

This was previously a Google Form, they've since migrated into a dashboard with actual confirmation.

Thanks for sharing, filed - now part of the elusive club.

Note: they ask where you live. I get the feeling this is more of a request than a choice

Agree with the peer, this is absurd. Legitimately worse than only having the original dark pattern.

ridiculous to have to find this buried page and go through an email confirmation

I always make sure to thank chatgpt when it does a good job. I like to think I'm helping to improve the model :)

I do it on the off chance that the ChatGPT history is used to train future AGI's, and want to ensure that my descendants are at least considered well behaved pets if nothing else.

This too.

Roko does the same thing, but for different reasons

I was waiting for this.

Google showed that providing an input field in which people can put their questions and then display ads next to the answers is one of the best business models ever. If not the best business model ever.

Google was free, open and without ads in the beginning. It is just too tempting to become the Google of the AI era to not try and replicate this story.

Microsoft is trying it too, but for some reason, they do not provide the simple, clean user interface which made Google successful. I always wondered, why they didn't do that with Bing search. And now with Bing copilot they chosen the same strange route of having a colorful and somewhat confusing interface.

Let's see how OpenAI does.

Google.com's "clean UI" differentiator gave it a first-mover advantage compared to the average 90s search engine, but that's as far as the UI impact goes.

From 2007 onwards, Google has maintained its dominance through decidedly non-search channels: the primacy of Youtube, the Chrome browser, Android OS, and of course, paying Apple billions to make Google its default search engine.

How do you explain that Bing has only 10% of the search market share on the desktop?

Windows has over 70% market share on the desktop. Windows comes with Edge as the default browser and Edge has Bing as the default search engine.

But Edge only has 12% of the Desktop browser market share.

It looks like 6 out of 7 windows users switch to Chrome for some reason. If not for the cleaner interface - why?

>It looks like 6 out of 7 windows users switch to Chrome for some reason. If not for the cleaner interface - why?

They don't switch to Chrome. They're already using Chrome. And odds are, they probably have been since the early 2010s, if not earlier---long before Edge was a thing.

When they get a new computer, they install Chrome because they're already in that ecosystem: bookmarks, saved passwords, customization, Google accounts, familiarity. They won't suddenly use Edge because it has none of that.

I literally only use chrome on windows because of the cleaner interface.

Because when IE was utter garbage, Chrome had better performance, an ecosystem that included GMail, and also stored our passwords and bookmarks. Chrome also eventually allowed you to conduct google searches directly from the address bar. People use what they are comfortable with, and all the functionality built into Chrome by google is a HUGE sunk cost to bypass. The only person I know who uses Edge and Bing regularly does it to earn gift cards from Microsoft.

> How do you explain that Bing has only 10% of the search market share on the desktop?

It's because downloading Chrome is the first piece of advice anyone receives when setting up a new computer, Windows or Mac.

Advice from whom?

Grandchildren, nephews, nieces, etc.

Google shills, presumably.

Also, Gmail.

People forget just how good gmail was. It wasn't just a geek status thing, although getting into the beta certainly helped, it was more that every other option was absolutely inodiated with spam, and, worse still, most had very poor tooling available for managing email in general.

Back in the dark ages when I was using Yahoo, I was receiving plenty of email, and about 80% of it was spam, switched to gmail and never had that problem again.

It was, honestly, quite an undertaking to de-google myself, because I'd been at it so long.

Gmail's #1 selling point, as I remember it, was a much more generous amount of storage, for free. I think Hotmail had like 20MB of free email storage, Gmail had one gigabyte.

True, storage was also a thing, I remember using some hacked together GmailFS program, where you could split a large file among a bunch of email attachments that were sent to yourself and tagged, then you could download and reassemble those files anywhere.

I used it to store FLV music videos I found, it was... the okayest form of cloud storage.

Indeed, storage! I think it grew for each person you invited, too, or some scheme like that...

I think it depends a lot on usage, I have a "real email" and consider my gmail address as "disposable", I use it to sign up for websites etc, whenever a form ask me to include an email address I will use my gmail address - the only time I ever log in to it is when I need to reset a password or similar.

It is full of spam.

Yes, that's a big one. Google searches are no longer anonymous now because people are mostly in a logged-in state on their browser thanks to Gmail or YouTube.

> We’ve also introduced additional content safeguards for this experience, such as blocking prompts and generations in a wider range of categories.

Oh good, just what wasn't needed.

I get that this sucks but what's the alternative? It would take legislation that will never pass to eliminate OpenAI's liability for the content that their model spits out and they're already under a microscope by regulators and the public.

This doesn't have anything to do with their liability or regulations, it's what their paying customers want. A company that pays them to provide a chatbot for their business doesn't want the chatbot to say controversial things or tell people to commit suicide.

Actually if it's about customers not paying them if they spit out "kill yourself" content, then I 100% understand that.

These are businesses. Those servers cost money. That compute costs even more. The AI experts, (real AI experts by the way, not tensorflow and pytorch monkeys), cost big money as well. Someone better be paying for all that.

So if the people who want offensive content are willing to pony up the dollars, then great. I've got no problem with giving them what they want. But if they want Barbie or Cocomelon to pay for the offensive content, then yeah, they can be safely ignored. Block as much as you like. (Or rather, block as much as Barbie wants blocked. Which is probably more than you like, but they're paying you handsomely for it.)

I may just be a pytorch monkey, but I cost more than a training server.

The alternative is running models locally

This would be ideal, but what percentage of the world's population can afford the hardware required for that?

The alternative is to stop giving a pass to censorship obviously?

You obviously don't understand what liability means.

Alternative would be to have model itself aware of sensitive topics and and meaningfully engage with questions touching such topics with that awareness.

It might be a while till that is feasible, though. Until then, "content safeguards" will continue to feel like overreaching, artificial stonewalls scattered across otherwise kinda-consistent space.

Is this for no-account users, or for everyone, including paid users?

In the past, I thought that search engines censored content because of the advertiser's demands, but now that AI search switched to the paid model I think the censorship is purely ideological because these companies are based in San Francisco.

I think search engines block content because it is what users want.

In the early internet, you'd almost always end up with a few pictures of naked girls in the search results, and many pages showed ads for porn sites.

That's just not a good experience if you are at work and trying to get stuff done.

Search engines have gotten much better, and in most cases explicit results are now only shown when you actively look for them.

The censorship is about not getting sued or regulated

So if you picked up the building and put it elsewhere it'd produce the same product but different politics?

This is one hell of a loss leader to keep people from using competitors, especially since it appears that it’s not being used to upsell to a paid offering.

This can’t be sustainable even with all the inference optimizations in the world.

> We may use what you provide to ChatGPT to improve our models for everyone.

I believe OpenAI has elected to eat the compute cost in order to teach the model faster to stay ahead of competitors. How much are you willing to pay as the robot gets better faster? Everyone fighting over steepness of the hockey stick trajectory.

Anyone can buy GPUs, you can’t buy human attribution training. You need human prompt and response cycle for that.

I agree, GPT-4 is still the undisputed king in LLMs, it's been a wee bit more than a year since it came out. I'm sure that the quality of GPT-5 will depend more on a carefully curated training set than just an increase in their dataset size, and I think they're really good at that.

Also, very few people know as much as sama about making startups grow, so ...

PS. I'm not even a fan of "Open"AI, but it is what it is.

It's more of a disputed king these days, lots of benchmarks show Claude Opus ahead and a fair few people do prefer it.

Yup, Claude is my go-to now. Both models seem equally "intelligent" but Claude has a better sense of when to be terse or verbose + the responses seem more natural. I still use GPT4 when I need code executed as Anthropic hasn't implemented that yet (though this "feature" can be annoying as some prompts to the GPT4 web interface result in code being executed when I just wanted it to be displayed).

No joy:-(

Unfortunately, Claude.ai is only available in certain regions right now. We're working hard to expand to other regions soon.

ironically the API is available most everywhere, and it has an ok-ish web interface (they're really heavy on prompt engineering it seems though)

I think OpenAI are running scared of Anthropic (who are moving way faster than they are). The last half dozen things they have said all seem to point to this.

"New model coming soonish" (Sure, so where is it?)

"GPT-4 kind of sucks" (Altman seemed to like it better before Athropic beat it)

"[To Lex Fridman:] We don't want our next model to be shockingly good" (Really ?)

"Microsoft/OpenAI building $100B StarGate supercomputer" (Convenient timing for rumor, after Anthopic's partner Amazon already announced $100B+ plans)

"ChatGPT is free" (Thanks Claude!)

> "[To Lex Fridman:] We don't want our next model to be shockingly good" (Really ?)

Yes, really.

They're strongly influenced by Yudkowsky constantly telling everyone who will listen that we only get one chance to make a friendly ASI, and that we don't have the faintest idea what friendly even means yet.

While you may disagree with, perhaps even mock, Yudkowsky — FWIW, I am significantly more optimistic than Yudkowsky on several different axies of AI safety, so while his P(doom) is close to 1, mine is around 0.05 — this is consistent with their initial attempt to not release the GPT-2 weights at all, to experiment with RLHF in the first place, to red-team GPT-4 before release, asking for models at or above GPT-4 level to be restricted by law, and their "superalignment" project: https://openai.com/blog/introducing-superalignment

If OpenAI produces a model which "shocks" people with how good it is, that drives the exact race dynamics which they (and many of the people they respect) have repeatedly said would be bad.

influence from yudkowsky is surprising, considering if you've ever touched hpmor, you'd realize the dude is a moron.

re p(doom): the latest slatestarcodex[0] has a great little blurb about the difficulties of applying bayes to hard problems because there's too many different underlying priors which perturb the final number, so you end up fudging it until it matches your intuition.

[0] https://www.astralcodexten.com/p/practically-a-book-review-r...

I find it curious how many people severely downrate the intelligence of others: to even write a coherent text the length of HPMOR — regardless of how you feel about the plot points or if you regard it as derivative because of, well, being self-insert Marty Stu/Harry Potter fanfic[0] — requires one to be significantly higher than "moron", or even "normal", level.

Edit: I can't quite put this into a coherent form, but this vibes with Gell-Mann amnesia, with the way LLMs are dismissed, and with how G W Bush was seen (in the UK at least).

Ironically, a similar (though not identical) point about Bayes was made by one of the characters in HPMOR…

[0] Also the whole bit of HPMOR in Azkaban grated on me for some reason, and I also think Yudkowsky re-designed Dementors due to wildly failing to understand how depression works; however I'm now getting wildly side-tracked…

Oh, and in case you were wondering, my long-term habit of footnotes is somewhat inspired by a different fantasy author, one who put the "anthro" into "anthropomorphic personification of death". I miss Pratchett.

If you think that volume is a replacement for quality, you really need to read more. If you think volume is correlated with intelligence, you really need to read and write more.

In case you still don't believe me, you are welcome to hop onto your favorite fan fiction site, such as AO3 [0], and search for stories over [large k] words.

[0] https://archiveofourown.org/works/search?work_search%5Bquery...

That volume coherently. Obviously mere volume can be done by a literal monkey on a literal typewriter.

Also, I didn't say "correlated with intelligence", what I said was more of a cut-off threshold — asserting that one cannot be an actual moron given writing coherently on that scale is more of a boolean than a correlation.

I do need to write more (my own attempt at a novel has been stalled at "why can't I ever be satisfied with how I lead into the dramatic showdown!" for some years now, as none of my attempts have pleased me once written); but as for reading? Well, if you think my taste must result from insufficient breadth and depth, I must wonder what you think of Arthur C. Clarke, Francis Bacon, Larry Niven, Alexandre Dumas, Adrian Tchaikovsky, Neal Stephenson, Robert Heinlein, Alastair Reynolds, Isaac Asimov, Carl Jung, … I'm going to stop there rather than enumerate my whole library, but I can critique each in a different way without resorting to calling them playground insults, even the ones I dislike.

But I will say it was interesting to contrast Chris Hadfield's "An Astronaut's Guide to Life on Earth" with Richard Wiseman's "Shoot for the Moon" — or H. G. Wells with Jules Verne.

It's bad writing. It's objectively terrible writing in fact. Purely on a volume standpoint, the average novel is 100K words. The Brothers K is HALF the length.

If we ignore that it's a rewriting of JKR's 7 novel series, which gives it a certain amount of coherency, Yudkowsky violates almost every writing guideline in a bad way. In fact, I could probably write an infinitely long coherent essay describing the ways HPMOR violates a reader's mind. It would be easy given almost 700K source material.

But to point at some gaping holes, the plot has no pacing, and the entire story is a badly written self insert where the mc goes around and "fixes" JKR's plot-holes by writing themself into a corner.

The solution to this, is, of course, to write another 20K words of expecto patronum, dispel the plothole with more rationalist bullshit.

Your average fanfic is probably better written.

> It's bad writing. It's objectively terrible writing in fact. Purely on a volume standpoint, the average novel is 100K words. The Brothers K is HALF the length.

I infer you favour Blaise Pascal: "I'm sorry I wrote you such a long letter; I didn't have time to write a short one."

I always thought that was Mark Twain who said that, so I asked ChatGPT who initially told me to not self harm myself and that I’m not alone... But, yep, I have been misinformed my whole life. It was Pascual. Apparently the chat bot thinks I’m having an existential crisis over this but thank you for educating me.

yud's goal was to spread concepts and popularity about his rationality cult, and he arguably did that very successfully with HPMOR. i don't think he's an idiot if he successfully reached his goals, and is also currently basically providing himself a full time job (and employing many others too) doing exactly what he wants to do with no clear profit motive to the people who fund him

his writing may be a little cringe at times and not anywhere near the prestige of "real writers" but it's perfectly entertaining for his intended audience

It's interesting that you're slinging around moron accusations at Yudkowsky seemingly unaware that slatestarcodex thinks very highly of his intelligence and Scott's blog is downwind of Yudkowsky.

How is Anthropic moving so fast if it took them almost a year to produce a better model? And they have probably 1% of the market right now.

Anthropic as a company only was created after GPT-3 (Dario Amodei's name is even on the GPT-3 paper).

So, in same time OpenAI have gone from GPT-3 to GPT-4, Anthropic have gone from startup to Claude-1 to Claude-2 to Claude-3 which beats GPT-4 !

It's not just Anthropic having three releases in time it took OpenAI to have one, but also that they did so from a standing start in terms of developers, infrastructure, training data, etc. OpenAI had everything in place as they continued from GPT-3 to GPT-4.

I pay for both of them and I keep finding myself coming back to GPT-4. Not only do I think the UI is vastly superior, I have not experienced a significant difference in quality of output between the two. I regularly ask both of them the same question and respond with follow-ups to both of them.

I had a funny thought when Anthropic was still a new startup. I was browsing their careers page and noticed:

1. They state upfront the salary expectations for all their positions 2. The salaries offered are quite high

and I immediately decided this was probably the company to bet on, just by virtue of them probably being able to attract and retain the best talent, and thus engineer the best product.

I contrasted it with so many startups I've seen and worked at that try their damnedest to underpay everyone, thus their engineers were mediocre and their product was built like trash.

Agreed with the spirit of this post. OpenAI also pays very well though and has super high caliber talent (from what I can see from friends who have joined and other online anecdotes).

Depending on the state, it is now a legal requirement for salaries posted up front. This is a requirement in California, Colorado and NY.

But yes it also helps attract talent.

Every paying customer Anthropic gains is a paying customer that OpenAI loses. The early adopters of OpenAI are the ones now becoming early adopters of Claude 3. Also 200k context window is a big deal .

I don't completely disagree with you but personally, Claude 3 doesn't seem like a big enough upgrade to get me to switch yet.

I have also personally found that minimizing the context window gives the best results for what I want. It seems like extra context hurts as often as it helps.

As much as I hate to admit it too but there is a small part of me that feels like chatGPT4 is my friend and giving up access to my friend to save $20 a month is unthinkable. That is why Claude needs to be quite a big upgrade to get me to pay for both for a time.

I kind of feel the same way. I have a lot of chats in Chatgpt .. I have also been using it as a sort of diary of all my work . The Chatgpt app with voice is my personal historian!

However once I figure out a way to download all my chats from Chatgpt, I think Claude' 200k context window may entice me to rethink my Chatgpt subscription.

I've only been using GPT via Bing CoPilot... How does the history work in the ChatGPT app? Is it just that old conversations are stored, or are they all part of the context (up to limit)?

It's going to be GPT-3.5turbo not GPT4. As others have mentioned, binggpt/bard are also free.

These smaller models are a relatively cheap to run, especially in high batches.

I'm sure there will also be aggressive rate limits.

they regulary and often downgrade their gpts. gpt4 now is about as good as gpt3.5 was in the beginning.

like 6-7 months ago there was a gpt4 version that was really good, it could understand context and stuff extremly well, but it just went downhill from there. i wont pay for current chatgpt 4 anymore

While I agree that GPT4 (in the web app) is not as good as it used to be, I don't think it is anywhere near GPT3.5 level. There are many things web app GPT4 can do that GPT3.5 couldn't do at ChatGPT's release (or now afaik).

One thing I really dislike about hosted models is how opaque that behavior is. As a user I should never be guessing if they've reduced a model's capabilities to save on compute for example.

This is why I'm excited for the growth of local model capabilities. I can much more reasonably expect that the model has not degraded and that it is using the full hardware capabilities it has been granted.

If the value of user interactions exceeds the cost of compute, then this is an easy decision.

They apparently constrained this publicly available version, with no gpt-4 or dall-e, and a more limited range of topics and uses than if you sign up.

They do explicitly recommend upgrading:

  We’ve also introduced additional content safeguards for this experience, such as blocking prompts and generations in a wider range of categories.

  There are many benefits to creating an account including the ability to save and review your chat history, share chats, and unlock additional features like voice conversations and custom instructions.

  For anyone that has been curious about AI’s potential but didn’t want to go through the steps to set-up an account, start using ChatGPT today.

Bing/Copilot works without an account too, it's just very limited and it will ask you to sign in way too often for it to be useful, openai will probably do the same thing.

There is still a huge volume of people who haven't used a large language model, and haven't had their own "aha" moment. Getting your core functionality available on the front door is good marketing and they have the funding to burn GPT-3.5 tokens.

I know what you mean, but in terms of phrasing, one might even say they have a way of cheaply creating more GPT-3.5 tokens if they ever run out :)

It seems most of these companies are increasingly using synthetic data (use one generation of LLM to generate specific types of training data for the next) rather than just looking for more/better human generated data.

my interpretation: MAU are going down rapidly and they need a KPI to counter that

This. It was a nice toy, but people moved on when they realized it couldnt do useful valuable work

I really don’t understand this viewpoint. I use ChatGPT almost daily to help with my professional work. I regularly ask it to do things that I know how to do and have done many times, but don’t want to write from scratch. That coupled with things I know how to do or know are possible, but don’t wanna have to go read all the documentation to remember the parameters to pass into a CLI tool or similar.

I agree with this as a professional but aren’t many students likely still using LLMs?

Anecdotally, I believe a lot of college administrators have realized that Google Docs / etc with tracked changes is difficult for students to fake, relatively easy to "sniff test" for plausibility, and helps protect students against false positives from AI/plagiarism detectors [ugh]. Maybe the gig is up.

Also anecdotally: despite the embarrassing scam of "AI detectors," many kids did not actually get away with cheating via ChatGPT last year. Issues like fictional citations, "as a large language model I cannot," and making up facts affect high school English papers just like legal filings or scientific articles about rats. And unlike scientific peer review, high school teachers usually read things closely and notice when something isn't right. You don't need an AI detector to be suspicious when an impeccably well-written essay discusses events in Great Expectations that didn't happen.

I've had a feeling for a while that they've hit a plateau in terms of how much "intelligence" they can squeeze from all the text in the world. Many openai and other researchers have publicly claimed that they need better data for better models. This to me feels like a desperate attempt to collect more data than they already are to improve their models.

Your feelings are dead wrong.

First, they have not hit a plateau. If you're in any ways involved with AI research you'd know that there is an insane amount of low hanging fruit in terms of data (synthetic and real world), architectural improvements, loss function improvements and scaling.

I also doubt their main motivation is collecting more data. Their motivation is directly competing with Google on their search business. The early success of Perplexity has shown that an answer engine built on top of LLMs is an improvement over a list of ranked web pages. OpenAI would be stupid to not go after that market, especially given the kind of mind-share they already have. It's clear that Google is stuck facing the innovator's dilemma.

And this is not speculation. Sama said (in Lex's podcast) that taking on Google is something that he's very interested in. Coincidentally it also aligns with OpenAIs mission of making this tech benefit everyone.

I hope I am wrong. All I've seen in the past year are incrementally better models even from the top labs. Open source models are barely usable.

We may use what you provide to ChatGPT to improve our models for everyone. If you’d like, you can turn this off through your Settings.

Please correct me if I'm wrong, but I do not think this can be GDPR compliant: there's a good chance people might enter personal information and in that case, OpenAI cannot just use said data without explicit consent for their own purposes. And the keyword here is "explicit" - just saying "by using ChatGPT, you automatically agree to your data being used by us - just don't use it if you don't agree, or turn it off in the settings" does not work.

It does not look like this is rolled out to Europe, or at least not broadly. (Tested using a VPN to France)

I believe OpenAI is doing this because they are scared of Microsoft.

More specifically…. Bing Chat is free and you can converse with it without using an account for a few prompts at a time.

Just go into Incognito into your browser if you run into any limits on Bing Chat.

Bing Chat uses OpenAI tech but OpenAI doesn’t make money from it. So OpenAI is probably worried people will use Bing more. Interesting kind of relationship they got going on with Microsoft. They need to provide Microsoft with the tech to access Microsoft data centers but this leaves them the risk of Microsoft overtaking them in the AI space with their own tech itself.

Fascinating

Makes sense, since Bard + Bing are free and don't require sign in

OpenAI has a big problem, the only moat on an LLM is the quality and inference speed. Which is quickly getting commoditized.

The comment above seems to be interpreting OpenAI’s goals in a way that is not consistent with its charter.

Structurally, for OpenAI, capped profit is a means not the end. If the capped profit part of OpenAI is not acting in alignment with its mission, it is the OpenAI board’s responsibility to rein it in.

—

https://openai.com/charter

OpenAI’s mission is to ensure that artificial general intelligence (AGI)—by which we mean highly autonomous systems that outperform humans at most economically valuable work—benefits all of humanity. We will attempt to directly build safe and beneficial AGI, but will also consider our mission fulfilled if our work aids others to achieve this outcome. To that end, we commit to the following principles:

Broadly distributed benefits

…

Long-term safety

…

Technical leadership

…

Cooperative orientation

Turboderp released a custom quantized version of dbrx-base and instruct a few days after the official model landed. https://huggingface.co/turboderp/dbrx-base-exl2

You are right, LLMs weights are fast getting commoditized. As fast as compute is scaling now, I don't think it can continue forever, so the huge advantages big players have is going to get wiped. I'm sure they will always be able to deliver at scale, but a super GPT4 performing model available on your local (or small player type) hardware in the next few years seems more likely than a few big mega-players and mega-models.

Is it? I thought this would be the case by now but GPT-4 is still in the lead (and released a while ago). How many generations in a row does OpenAI have to dominate before we revisit the "there is no moat" idea?

i've been a paid consumer of chatgpt - 4 for a while and i've been recently using gemini more and more, its free and while not as good as gpt-4, it is good enough where paying for gpt-4 doesn't seem worth it anymore.

Anthropic Opus is better than GPT 4 on many (most?) tasks.

Not only better as measured by benchmarks, but also better preferred by humans.

https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboar...

The most surprising thing to me is that Opus is only slightly in the lead.

I was feeding multiple python and c# coding challenges / questions to both and Opus blew GPT4 out of the water on every single task. Didn’t matter if I was giving them 50 lines or 5,000 Opus would consistently give working/correct solutions while GPT4 preferred to answer with pseudo code, half complete code with ‘do the thing here’ comments, or would just tell me that it’s too complicated.

Another data point, I definitely find Opus better for coding, but not by much. The problems I give them are generally short (<= 100 lines) and well-defined so any advantage Opus has in larger contexts won't be apparent to me. They're also generally novel problems but NOT particularly challenging (anyone with a BS CS should be able to solve them in < 1hr).

I have them working with mostly C++ and Clojure, a bit of Python, and Vimscript every once in a while. Both models are much better at Python and fairly bad at Vimscript. Clojure failure cases are mostly from invented functions and being bad at modifying existing code. I can't pick out a strong pattern in how they fail with C++, but there have been a few times where GPT4 ends up looping between the same couple unworkable solutions (maybe this indicates a poor understanding of prior context?).

Spot on. People need to say what they are actually using the models for and not just "coding".

I mostly use it to make react/javascript front ends to a python/fastapi backend and chatGPT4 is great at that.

I tried to write a piece of music though in the old Csound programming language and it barely even works.

It will be interesting to see how the context plays out because I have noticed that I can often give it extra context that I think will be helpful but end up causing it to go down a wrong path. I might even say my best results have been from the most precise instructions inside the smallest possible context.

It's because LMSYS is an aggregate elo across a range of different tasks. Individually in some very important areas, Claude Opus may be better than GPT-4 by 50-100 elo points which is quite a lot. However there are specific domains where GPT-4 has the advantage because it's been fine tuned based off a lot of existing usage. So weak points around logic puzzles or specific instructions don't bring down its elo whereas Claude Opus doesn't have this advantage yet. I believe Opus's eventual elo, after all these little areas of weakness are fine tuned, will be something like 1300.

Yeah GPT is incredibly lazy, ironically 3 is far better at not being lazy than 4.

I guess you benchmarked via API? I've heard even the datestamped models have been nerfed from time to time..

Same here. I pay for both GPT-4 and Opus. Opus is miles ahead on software development work for me.

Agreed. Opus definitely has taken most of my usage as of late.

We don't talk about Google "not having a moat" since their patents on PageRank expired.

All the talk of OpenAI's moats (or lack thereof) since the memo, seems like humans being stochastic parrots.

Google has a data moat on traditional search. Its really hard to train a traditional search system without google scale data. LLMs effectively side stepped that whole issue

Google's special sauce, is the data of all user's responses to the things Google's algorithm shows them.

This description also works for ChatGPT, what it has which others do not (at anything close to the same scale).

Google's search engine is a well understood bit of matrix multiplication on a scale many others can easily afford to replicate.

This description also works for GPT-3.

No becuause ChatGPT was trained on the whole internet using a new algorithm. You needed perfect tagged data for the old algorithm (page rank). LLMs are a fundamental leap forward

The GPT algorithm isn't a secret.

The RLFH data — users giving a thumbs up or down, or regenerating a response, or even getting detectably angry in their subsequent messages — is.

PageRank isn't a secret.

Google's version of RLFH — which link does a user click on, do they ignore the results without clicking any, do they write several related queries before clicking a link, do they return to the search results soon after clicking a link — are also secret.

That the Transformer model is a breakthrough doesn't make it a moat; that the Transformer model is public doesn't mean people using it don't have a moat.

Hence why I'm criticising the use of "moat" as mere parroting.

Google's acquisitions are its moat. Google Search is only one small part of that

Google wouldn't be the company it is if it didn't also control Youtube, Gmail, Android and Chrome.

Yep, Groq is one example

Great example: can't scale enough to remove their 30 requests per minute (!) rate limit and enable billing, barely meets 3.5 levels of intelligence, etc;

People don't seem to understand that scaling out LLMs efficiently is it's own art that OpenAI is probably learning lessons on faster than anyone else

You're thinking of Grok. Not Groq. Blame Elon for this naming catastrophe.

Lol what? No, I'm thinking of Groq.

I love Groq though, every time someone hypes up their project using it you instantly know they have no real usage whatsoever.

Even the most "toyish" toy project doesn't fit in their rate limits for anything more than personal use.

Trying this on my phone really sucked. After submitting my question, about half the screen was covered with a sort of cookie consent popup with no visible way to dismiss it, and only a few pixels of the response were barely visible. Ridiculous.

There’s no such thing as a free lunch.

Yup, in exchange they get their models trained by even more people.

Kind of win-win situation.

> Starting today, you can use ChatGPT instantly, without needing to sign-up

It’s actually been up since at least Thursday of last week, maybe longer

I wanted to show ChatGPT to someone who hadn’t used it before, they went to chat.openai.com and were able to use it without creating an account

Wait, so this is not an April Fools joke?

OpenAI starts looking slightly desperate.

No surprise here.

I'd settle for just not logging me out every day.

Mine doesn’t? Weeks on end

Failure: missing noscript/basic (x)html interop

April's fool?

I'll never understand companies making genuine product announcements on April 1st, the most famous example of this has to be Gmail. I mean, really? You just have to do it on this day? Couldn't just do it yesterday or tomorrow, or any of the 364 other days in the year?

The gmail one is a classic though. It was an excellent way to bring awareness. It seems to have worked out for them too

Well, the "beta" tag turned out to be a joke.

On the other hand it could be a great way to float wild product ideas and see if there’s interest.

I’m surprised no companies try that.

You can do that directly with customers without doing a public announcement and potentially confusing people. It's a bad idea.

Companies coming up with outlandish claims on April 1st can look fun or enticing, and that's great PR. Otherwise, when a company does it because it's April 1st or worse, the CEO is the only one who thinks it's fun...it's just cringe.

That's cool: it means that I can now log off OpenAI and that means that under GDPR OpenAI now has no right to keep any of the data I'm feeding to ChatGPT.

The ability to use them anonymously used to be my only criteria for interacting with an llm operated by a third party.

My anonymity for your free training data seemed a reasonable barter.

It’s the reason phind is really the only one I use at all.

But gpt4’s parent company has proven itself to thinking it is above the law, or even good faith morality, that I refuse to provide them any more data than the shit they already stole from me and my fellow creatives.

May thy api chip and shatter.

This is the way.

Now let's see the results of having one of these bots debate another one, a verbal version of Robot Wars (BattleBots on the western side of the Atlantic). Gentlemen, prepare your chatbots!

I think more and more people are slowly realizing that LLMs aren't products in themselves. At the end of the day, you need to do something like Midjourney or Copilot where there's some value generation. Your business can't just be "throw more GPUs at more data" because the guy down the block can do the same thing. OpenAI never had any moat, and it's a bit telling that as early as last year, everyone on HN acted as if they did.

LLMs are like touch screens: technically interesting and with great upside potential. But until the iPhone, multi-touch, the app ecosystem comes along, they'll remain an interesting technological advancement (nothing more, nothing less).

What I'm also noticing is that very little effort (and money) is actually spent on value generation, and most is spent on pure bloat. LangChain raising $25m for what is essentially a library for string manipulation is crazy. (N.B. I'm not solely calling out LangChain here, there are dozens of startups that have raised hundreds of millions on what is essentially AI tooling/bloat.) We need apps--real, actual apps--that leverage LLMs where it matters: natural-language interaction. Not pipelines upon pipelines upon pipelines or larger and larger models that are fun to chat with but actually do basically nothing.

The need for concrete products is acute. I'm in the martech space and it's ridiculous how many big-name software vendors have done nothing besides slap an "AI-powered" label on their traditional products (CRM + email marketing and API-connector type apps).

Tried talking to the reps about the AI features. All were very cagey, suggesting that they didn't really know and were parroting buzzwordy benefits because the higher-ups told them to. So far, the "AI revolution" seems to have just led to higher price tiers.

> The need for concrete products is acute.

This feels very crypto-like in this regard for me. Businesses seem to be struggling with what we're all struggling with, LLMs are untrustworthy.

Business' needs are clear. Handle payroll, handle data analysis, conduct sentiment analysis of our brand across social media, identify problems. Do this effectively and at a far lower price than what a human would agree to work for.

AI isn't moving the needle much here, which is why GenAI "solutions" are underwhelming things like support chatbots, writing assistance and auto-generated stock art for Facebook ads.

There's nothing wrong with doing that a service but it falls well short of the fantastical visions that Microsoft, OAI and SV at large have portrayed.

The LangChain fundraise was a year ago, the same time the ChatGPT API was released and there was much more optimism about generative AI.

Things have cooled a bit more nowadays as the cost economics became more real. I wouldn’t expect another LangChain level fund raise.

I agree with your points about LLMs being like touch screens but what kind of natural language interaction apps are you talking about? Might be that some startup has attempted something similar already.

I'm a different person but the key for me is a wrapper than can handle uncertainty and ask questions for clarification.

Currently I don't find LLMs tremendously useful because I have to be extremely specific with what I want, which IMO is closer to writing code than the magical promise of turning ideas into products.

An app that I can just talk into and say, "I want to do X" and it can start building X while asking me questions for functional requirements, edge cases, UI/UX, that's a killer app.

It's also an app which could actually decimate many SWE jobs, so, I should be careful what I wish for.

80% of people just want the ability to write a sentence and have it run the correct SQL on their data warehouse. That's the biggest use of LLMs I see in the enterprise right now.

I think you are right -- eventually we will see an ecosystem pop up that uses LLMs in unique ways that are only possible with LLMs.

But in the meantime people are using LLMs to shortcut what used to be long processes done by humans, and frankly, there is a lot of value in that.

> 80% of people just want the ability to write a sentence and have it run the correct SQL on their data warehouse.

Even this is IMO too ambitious (due to edge cases, and multi-shot prompting being basically required, which kind of defeats the purpose). Right now, I'm working on a product that can just do simple OS stuff (and even that is quite challenging); e.g. "replace all instances of `David` with `Bob` in all text files in this folder".

Reminds me of the crypto bubble. All the defi shitcoins giving yields to each other while no shitcoin actually generated real value or had real demand. Now everyone building LLM libraries and systems but very few LLM products yield real value

There's a big difference cost-wise: anybody can deploy an Ethereum fork, mine the resultant shitcoin very cheaply and start a Discord for marketing purposes. That's the whole product right there.

The same is not true of AI projects which require a lot more upfront investment.

All of the 25 comments so far are missing the point.

Surprisingly, YC's greater opinion is still "ChatGPT is a useless stochastic parrot." probably because they are too cheap to fork over $20 to try GPT4.

Yes, GPT3.5 is mostly a toy. Yes, it hallucinates a non-trivial amount of time. But IMO, GPT4 is in a completely different class, and has almost entirely replaced search for me.

If OpenAI really wants ChatGPT to challenge search, it has to be free and accessible without requiring a sign up.

I very rarely use any search engine now. Really I only use search when I'm looking for reddit threads or a specific place in Google Maps.

All of my other queries: how things work, history, how to setup my wacom, unit conversions, calculating mortgage, explain stdlib with examples, and so on... All of that goes to ChatGPT. It's a million times faster and more efficient than scrolling through endless SEO blog spam and overly verbose Medium articles.

This update makes ChatGPT3.5 available without sign up, not ChatGPT 4. But if/when ChatGPT4 becomes available without sign up, I have no doubt the rest of the population will experience the same lightbulb moment I did, and it will replace search for them as well.

> probably because they are to cheap to fork over $20 to try GPT4.

Or because GPT 3.5 was hyped to the skies by all and sundry, and those who were convinced enough to use it still found it lacking. Many like you are now saying "oh yeah GPT 3.5 was awful, but this really is the future".

Not everybody wants to have the quality of their work dependent on the whims of an OAI product manager. If GPT-4 is as good as claimed, then it will find its way into my workflow. For now, AI claims are treated as fiction unless accompanied with a JSFiddle-style code example...too much snake oil in the water to do otherwise.

Bard already integrates with Google Search. These companies aren't competing to be the best; just the most good enough in existing form factors.

> mostly ... but

I have been busy on different matters:

in absence (I presume) of a model of their model of reasoning,

has some metrics, some measurement, be produced to understand the deep reliability (e.g. absence of hallucination, logical consistency etc.) of LLMs?

It is bizarre to me that your comment was downvoted, it is incredibly insightful. Google is very worried about this right now and they should be.

I bet they’re doing this to preemptively get around any future laws that prevent young people from signing up.

No sign up required means no age verification either.

Why would young people be forbidden from using LLMs in the future?

Age-inappropriate responses, by whatever that means to whoever sets the rules.

Social media has been the boogeyman for almost a decade now, and (red) states are at least sending out trial balloons regarding banning minors from accessing social media[1]. Prior to that, it was porn, and now we have age verification required by law.

I'd argue that AI is a much bigger boogeyman than Instagram/Tiktok/Pornhub ever was.

[1] No judgement of whether this is a good idea or not; in some sense it probably is; but I feel the current discourse is reactionary/political and not really about actual people's actual well being.

> No sign up required means no age verification either.

That is an absurd conclusion to speculatively leap to, and wrong. Typically age-gating laws are written so that it doesn't matter if you have to create an account or not. Porn has always been the most common case for this kind of legislation. Most people don't create accounts to watch porn, and most porn is free without needing to sign up. The jurisdictions that require age verification still apply to porn sites where you don't need to sign up.

Reply on news.ycombinator.com