12 ms·

Local maxima go BRRRRRRR AI will get there eventually, but this current paradigm seems increasingly only useful for spam and shitty clip art. Even so, everyone

Local maxima go BRRRRRRR AI will get there eventually, but this current paradigm seems increasingly only useful for spam and shitty clip art. Even so, everyone is throwing absurd amounts of investment capital at it in the hopes that something useful will happen. It's a pretty clear depiction of the investor class being so detached from the technical reality of what they're investing in that they just sit around lighting billions on fire thinking that they're getting richer instead of poorer. Go ahead and buy more H100s though...

simonw 2y ago
I've been finding this stuff genuinely useful for two years now, across Copilot and ChatGPT and Claude 3 Opus and similar tools. Either I'm a dimwit, easily conned by hype and shiny tools to the point that I can imagine benefits for two years that simply aren't there... or there's something to them.
- aimazon 2y ago
  I don’t think you’re a dimwit but I read your post[1] with an example and I am curious to whether you feel you’re losing something by telling the LLM it’s wrong and to try again, rather than going through the exploratory/iterative learning process yourself. For example, would you have known to ask about GeoJSON if you had not come across and learned about it pre-LLM? More succinctly: do you feel you’re learning more or less or an equal amount when using LLMs? [1] https://simonwillison.net/2024/Mar/22/claude-and-chatgpt-case-study/ https://simonwillison.net/2024/Mar/22/claude-and-chatgpt-cas...
  - noduerme 2y ago
    You seem to be writing from a lost world when people enjoyed coding because they enjoyed learning new things with every project. I think that's sadly something only enjoyed at work by a very small minority these days.
    - simonw 2y ago
      I very much aim to learn something new with every project. The thing I like most about working with LLMs is that they let me take on MORE projects!
    - eru 2y ago
      I don't think those 'good old days' ever existed. Or rather, the situation hasn't really gotten worse.
      - noduerme 2y ago
        I probably wasn't clear I still live in that world, both professionally and in personal projects. But my perception is that the vast majority of devs and engineers do not.
  - simonw 2y ago
    Absolutely I could have learned more from that particular project if I'd spent more time with it rather than getting the LLM to do the work... but that's why I like it as an example: since it was effectively a distraction (a "side quest") the alternative wasn't learning more, it was not doing it at all (and learning nothing). I'm able to get really great results out of LLMs because I have 20+ years of experience helping me know what questions to ask of them. I do feel like my rate of learning has increased though, because I'm much more likely to try out a completely new technology when I know an LLM can flatten the learning curve for me a bit.
    - gtirloni 2y ago
      > I do feel like my rate of learning has increased It's the same for me. Lots of experience knowing what to ask for. It does a better job in summarizing knowledge and getting me a relatively coherent explanation. Much faster than using Google Search to find and connect the dots from dozens of pages. I just don't see much benefit in its reasoning and code assistance features besides basic stuff.
    - yawaramin 2y ago
      > I have 20+ years of experience helping me know what questions to ask of them. That seems rather unlikely given that LLMs didn't exist more than seven years ago, much less 20+.
      - Loocid 2y ago
        They mean they have 20 years experience in programming, which helps them ask good questions.
        simonw 2y ago
        Yeah, that's what I meant. With LLMs having deep domain knowledge in a field helps enormously in terms of prompting them as productively as possible.
        2y ago
  - erhaetherth 2y ago
    I was trying to fetch key-value pairs out of a database using PHP+PDO the other day and I knew there was a nice easy to do it but I couldn't remember how. Something about fetchAll, maybe PDO::FETCH_GROUP|PDO::FETCH_COLUMN.... what was it? So I asked a couple LLMs. They wrote out loops for me to format the data how I wanted. I could have copy-and-pasted that in and it would probably have worked. But I felt there was something better yet, so back to Google I go. It's `PDO::FETCH_KEY_PAIR`. It's built-in. But oddly kind of hard to find unless you know the right thing to search for, and "key pair" was not springing to my mind. Point is, if you just let the LLMs do your work you won't even find these better ways of doing things. And I'm quite afraid of what the LLMs are going to do to Google, Stackoverflow and documentation in general. In a couple users it'll be ungoogleable too.
    - fzeroracer 2y ago
      As noted LLMs can only give you what you ask for, but for a lot of problems what you ask for isn't what you need; it's two or three steps removed. And LLMs can't tell you that you're doing something wrong; unlike curmudgeonly users on SO or in various forums/channels. My gut feeling is that we're going to enter into a 'dark age' of coding where a lot of previously available resources are going to be ransacked and made hard to find in favor of big corporation owned LLMs. It's already having an extremely bad effect on search in general; we're potentially only a few fights away from sites like SO having users leave en masse. That's why I think having a strong network of engineers to talk with will become more important than ever, almost a return to the IRC days.
    - DrSiemer 2y ago
      Strong disagree. I have learned so many new things since I started using LLM's. Some of them I'm actually embarrassed to admit, because I should have known about them a decade ago. If you work in a small company and you are the most experienced developer, you don't often get feedback on how you can improve things. The trick is, quite simply: just ask. I regularly dump some code I wrote in a language model and then ask what can be done better. I would never do that in any online space, because first, I don't wait an answer maybe some day, I need an answer NOW. And second, I prefer to avoid being called a fool.
- anonzzzies 2y ago
  Nah, it seems a lot of people are using them wrong? OP seems to also fall in that category. I sit next to many people in pair programming situations and it’s quite weird to see very smart programmers using gpt or copilot; they are the ones shouting ‘stochastic parrot’ on forums and yet expect some kind of mind reading magic when using these tools. When I show that you can put a comment for your function, like you should do anyway, and it writes the function, it usually clicks. Many still find it ‘faster and easier’ to write the function themselves and that’s fine, but, like OP, doing things like entering ???? and generally expecting a useful result will not be optimal.
  - ern 2y ago
    You make an interesting point. I once worked with someone who was brilliant, but had gotten into the field via a math degree, and we were building an enterprise-software product. We tried pair-programming, and he cracked within a few minutes. He couldn't articulate his thought processes verbally. Sounds similar to the scenario you're describing.
- blibble 2y ago
  I wouldn't feel too bad, there's orders more magnitude more money put into this con vs. the last one (NFTs)
  - eru 2y ago
    Your counting of 'cons' seems pretty idiosyncratic. You know there have been lots of other technologies people have been working on in the meantime and concurrently? Some of them more questionable, some of them less.
- selcuka 2y ago
  I basically say the same when people ask me if the AI bubble is going to burst. While I agree that the current push for LLMs should be taken with a grain of salt, I don't think it's a bubble either. I have been finding it useful too, and I don't think I will stop using it because it will look less shiny in the future.
  - BlueTemplar 2y ago
    The dot com was a bubble that burst spectacularly enough, yet we still use the World Wide Web, and even some of those websites.
- tymscar 2y ago
  You are clearly not a dimwit, and you don’t only have way more experience than most people here, you also have some amazing projects under your belt. However I can’t help but notice that the vast majority of blogposts, talks, tweets, and basically everything else you do now is around LLMs. Do you not think that’s indicative of this being “hyped” and “shiny tools”?
  - ToValueFunfetti 2y ago
    If hype just means people are excited about it and talking about it, it should be a positive signal about the merit of the thing. I think people get the wrong idea by measuring P(X is good | people are excited about X) and finding it to be low. But P(X is good | people are not excited about X) is vastly lower, and 'hype' as here used is not really discrediting.
    - thaumasiotes 2y ago
      > I think people get the wrong idea by measuring P(X is good | people are excited about X) and finding it to be low. But P(X is good | people are not excited about X) is vastly lower Only if you're including things that don't exist. If you restrict yourself to things that people might conceivably talk about, that probability is actually very high, for the simple reason that bad things are exciting specifically because they are bad. Here are some boring things: - bread - sunlight - air - water - parents Note that starvation, darkness, suffocation, dehydration, and being orphaned are all much more exciting than their opposites.
      - ToValueFunfetti 2y ago
        Poor phrasing on my part; 'good' should have been 'worth talking about'.
  - tucnak 2y ago
    Deep learning is the most important thing to happen to philosophy since Socrates / Wittgenstein.
- thefaux 2y ago
  You are creating a false dichotomy. Very few people are saying there are no benefits, but many have reasonable concerns both about the current efficacy of these tools as well as the expectation of continued exponential growth which is driving much of the current hype. What if we have already passed the inflection point where the exponential growth transitions to an s-curve? That would mean that this technology on its own would only get marginally better than is today. Maybe 2, 4 or even 10x better than it is today, but not 100x or 1000x. To break those barriers, we would need further innovations beyond just throwing more gpus and training data at the problem. I personally am short on LLMs because I believe it is much more likely that we have already crossed the inflection point or will soon. Again, they are impressive but ultimately I think that LLMs will be at best a footnote in history if they are even remembered at all in a few hundred years. But of course I could be wrong.
  - simonw 2y ago
    I've been generally staying away from the whole "imagine what this stuff could do next!" side of the discourse. My personal opinion there is that if all research froze today it would still take us years to figure out all of the potential use-cases and applications for the models we have access to right now, and how best to apply them.
- vertis 2y ago
  No I'm with you. I don't use copilot though for undirected autocomplete, I use tools that allow me to give instructions and diff the results, Cursor and Aider are my current defaults, both with Claude 3 now[0]. Always looking for alternatives though. I think it's always going to be a YMMV experience with LLMs. I'm an extreme generalist with 23 years of experience so it compliments my strengths and weaknesses. I know how to program across 20 odd languages but most of them I need to look stuff up if I'm not using it frequently enough. Now though, don't really need to look stuff up. For me there are two groups, those that want to use LLMs and are pushing it forward and those that are would prefer that LLMs not exist and want it to be hype that goes away. Having followed the hype cycles of VR/AR and crypto I can feel a difference. Both of those felt like a solution in search of a problem. The true believers wanting it to be something like Ready Player One with VR (fun story, Oculus/Facebook actually handed out copies of the novel at Oculus Connect 2 in 2015). The hype around LLMs seems different; like applying a new solution to all the existing problems to see where it helps, and that's just with the current iteration. [0]: I'd prefer to be using local/open equivalents but the capabilities are still lacking.
mhh__ 2y ago
> spam The first 100 lines of any project is spam (usually 0 entropy), so AI is a nice win.
- eru 2y ago
  That depends on your language, and project.
- YurgenJurgensen 2y ago
  If the entropy is 0, you didn't need an LLM. If it's the same every time you can just copy and paste those 100 lines from GitHub or StackOverflow.
  - mhh__ 2y ago
    Why bother copying and pasting though, that's my whole point. Google is terrible in comparison, just let it write the code for you.
noduerme 2y ago
I think the nomenclature of calling generative AI an "AI" is just hype and that leads to disappointment. After all, it's much more exciting for investors than calling it the world's most expensive textual/audio/visual autocomplete. But that's really what this generatoon currently is. And it can feel like magic, because we're pattern-seeking and pattern-matching creatures, so something that seems to intuit the pattern we're looking for (even if imperfectly) can feel quite a bit like reasoning with another human. To be a bit less generous to humans, it's often the same as talking with another human, because we spend much of our lives writing boilerplate or making small talk or otherwise just kind of on autopilot and answering patterns with expected responses without seriously engaging our complex reasoning. (Do you have to really think about how to map-filter-reduce a dataset anymore? Or think seriously about how you really are when someone says "how are you"? We don't usually expect other people to, and most of the time we're satisfied by exchanging recognizable / somewhat coherent language patterns which we can count on each other to fill in the gaps of. But this is not the intelligence in human intelligence). The corporate hype machine works overtime to hype it as a solution, but that's just what they do. That's all spin. We see through it now, and we're getting to the point where people are asking the only important question about this tech, which is, how much will you pay for a nerfed autocomplete to do X?
- tylerchilds 2y ago
  “how much will you pay for a needed autocomplete to do X?” i think this is a super important for anyone playing in the LLM space to calculate. currently “AI” providers are selling electricity at a loss to demonstrate product “value”, so even while asking questions is “free” today, there’s actually a finite resource under the hood that needs to have the bill paid in the end. an estimate, but from historic trends, free bills come due in around 7 years
  - eru 2y ago
    Even _if_ we don't see our current 'AIs' become more intelligent, I am very certain that we will see the amount of electricity needed to produce 2024-state-of-the-art results drop dramatically over the next few years. People are only just figuring out how any of this works, and they are happy to get any reasonable results at all. They aren't really competing on costs like energy efficiency, yet. --- However I don't really think that 'if' will come to pass: I expect that we will still see lots of advances in the 'intelligence' of these models, so people will compete on these, and worry about electricity usage afterwards.
echelon 2y ago
> shitty clip art AI art is better than most humans, including most artists. You can take the outputs and immediately turn them into animation. Suddenly I, as a single individual, can easily make film content without figuring out set and lighting logistics or roping in dozens of people. This stuff is magic.
- ghaff 2y ago
  I was somewhat with you through the first sentence then it got a bit breathless. As someone with some graphic design experience but who isn’t very good and can’t really draw, I used Photoshop and it’s genAI capability recently to put together a book cover for an ebook. Would it win any awards? I’m pretty sure not but it’s more than adequate for its purpose and is almost certainly better than anything I could have come up with even if I used some CC artwork. ADDED: I also had something of a vision for what I was looking for and (I think) enough of an eye to fiddle with things and get to the point where I went "This isn't half bad."
  - echelon 2y ago
    Exactly! Editing and controllability are getting better almost every week.
- YurgenJurgensen 2y ago
  I've said it before and I'll say it again: The fact that you didn't think to include an example is telling. If you think AI 'art' is good, show me, and tell me why you think it's good.
  - zo1 2y ago
    Not OP. Conflating "art" (the "snobbish" definition if you will) with realistic depictions of imagined worlds, stories and settings is not helpful here. There are so many talented, imaginative people out there that have previously had Zero ability to manifest their thoughts into a visual form. AI has enabled that on so many levels and we should be grateful for it.
015a 2y ago
Yeah; its concerning how depressingly similar Sam's recent language is to, say, SBF's in his prime. Waxing poetically with faux-intellectualism about the sheer addressable market his business will capture (all of humanity) (and funded by the government) [1]. You can just listen to key players in the space, how their language has changed over the past year, and recognize that we're probably very near a local maxima / plateau / AI winter. Reality is: AI is startlingly expensive. Stupid, stupid expensive. Microsoft is building "Stargate", their $100B AI-focused supercomputer [2]. The money to build that isn't coming from selling AI services. Its coming from their old, boring, tried and true businesses; Windows, Office, Azure, and M365. Xbox is dying before our eyes. Their tried and true businesses make great money but aren't growing. Its up to AI to prove to investors that their P/E multiple is justified. [1] https://twitter.com/tsarnick/status/1789107043825262706 https://twitter.com/tsarnick/status/1789107043825262706 [2] https://www.tomsguide.com/ai/meet-stargate-the-dollar100-billion-ai-supercomputer-being-built-by-microsoft-and-openai https://www.tomsguide.com/ai/meet-stargate-the-dollar100-bil...
- viking123 2y ago
  I always mix SBF and Altman in my mind, and have to spend few seconds thinking which was which.. But IMO the current AI hype has some almost religious elements to it
dogcomplex 2y ago
It is honestly shocking how different people can be, even within programming. Been finding LLMs very useful for pair programming and low-effort spin-ups of projects I would just never do without the tools. Frankly, I'm waiting any day for an architecture to arise that takes lossy LLMs and has them error correct enough to produce 99.9999% reliable progress on simple tasks cheaply. This feels incredibly close, and if the unit of intelligence of each step of those is around GPT4 level? Jesus christ. Programming massively automated overnight. Any programmer that is this dismissive of these tools frankly... hasn't been using them right, and has no imagination - or a massive ego. This stuff is still far too early to make such judgements.
dartos 2y ago
I don’t really think AI will get there the way we’re going about it. Statistical models will always be limited by their data and how edge case filled reality is.
noobermin 2y ago
People are either misusing this meme or it makes no sense.
- CaptainFever 2y ago
  It's likely an ironic usage.
mondrian 2y ago
It's definitely useful, but in higher level chat use cases, not so much high-precision generation (yet). Rubber ducking, brainstorming, and search-like use cases are definitely a level above what we had.
n_ary 2y ago
> It's a pretty clear depiction of the investor class being so detached from the technical reality I think, their ROI model is much different and subtle. Few decades back, everything ran on-premises and costed ok ... Then a huge investment had been shovelled into fancy frameworks and pushing strange paradigms (microservices, distributed systems, monoliths whatever flavour works for the tech giants) so now people just put together whatever and needs massive _cloud_ to run basic things that used to run on few MHz devices. In the end, the ROI is NOT from these models, those are just gambles in case one actually becomes a king maker. The actual ROI is from the _cloud_ where innocent people will rent expensive hardware to try and utilise these models and funneling wealth to the actual investments(cloud operators, hardware vendors) which are already part of portfolio of these investors. Basically, $1MM invested in random shiny startup will inspire other shiny startups to also spend similar in race to become unicorn, all the while the cloud operators and GPU vendors are laughing to the bank preparing their shareholders(usually same investors) dividends.

simonw 2y ago

I've been finding this stuff genuinely useful for two years now, across Copilot and ChatGPT and Claude 3 Opus and similar tools. Either I'm a dimwit, easily conned by hype and shiny tools to the point that I can imagine benefits for two years that simply aren't there... or there's something to them.

aimazon 2y ago
I don’t think you’re a dimwit but I read your post[1] with an example and I am curious to whether you feel you’re losing something by telling the LLM it’s wrong and to try again, rather than going through the exploratory/iterative learning process yourself. For example, would you have known to ask about GeoJSON if you had not come across and learned about it pre-LLM? More succinctly: do you feel you’re learning more or less or an equal amount when using LLMs? [1] https://simonwillison.net/2024/Mar/22/claude-and-chatgpt-case-study/ https://simonwillison.net/2024/Mar/22/claude-and-chatgpt-cas...
- noduerme 2y ago
  You seem to be writing from a lost world when people enjoyed coding because they enjoyed learning new things with every project. I think that's sadly something only enjoyed at work by a very small minority these days.
  - simonw 2y ago
    I very much aim to learn something new with every project. The thing I like most about working with LLMs is that they let me take on MORE projects!
  - eru 2y ago
    I don't think those 'good old days' ever existed. Or rather, the situation hasn't really gotten worse.
    - noduerme 2y ago
      I probably wasn't clear I still live in that world, both professionally and in personal projects. But my perception is that the vast majority of devs and engineers do not.
- simonw 2y ago
  Absolutely I could have learned more from that particular project if I'd spent more time with it rather than getting the LLM to do the work... but that's why I like it as an example: since it was effectively a distraction (a "side quest") the alternative wasn't learning more, it was not doing it at all (and learning nothing). I'm able to get really great results out of LLMs because I have 20+ years of experience helping me know what questions to ask of them. I do feel like my rate of learning has increased though, because I'm much more likely to try out a completely new technology when I know an LLM can flatten the learning curve for me a bit.
  - gtirloni 2y ago
    > I do feel like my rate of learning has increased It's the same for me. Lots of experience knowing what to ask for. It does a better job in summarizing knowledge and getting me a relatively coherent explanation. Much faster than using Google Search to find and connect the dots from dozens of pages. I just don't see much benefit in its reasoning and code assistance features besides basic stuff.
  - yawaramin 2y ago
    > I have 20+ years of experience helping me know what questions to ask of them. That seems rather unlikely given that LLMs didn't exist more than seven years ago, much less 20+.
    - Loocid 2y ago
      They mean they have 20 years experience in programming, which helps them ask good questions.
      - simonw 2y ago
        Yeah, that's what I meant. With LLMs having deep domain knowledge in a field helps enormously in terms of prompting them as productively as possible.
        2y ago
- erhaetherth 2y ago
  I was trying to fetch key-value pairs out of a database using PHP+PDO the other day and I knew there was a nice easy to do it but I couldn't remember how. Something about fetchAll, maybe PDO::FETCH_GROUP|PDO::FETCH_COLUMN.... what was it? So I asked a couple LLMs. They wrote out loops for me to format the data how I wanted. I could have copy-and-pasted that in and it would probably have worked. But I felt there was something better yet, so back to Google I go. It's `PDO::FETCH_KEY_PAIR`. It's built-in. But oddly kind of hard to find unless you know the right thing to search for, and "key pair" was not springing to my mind. Point is, if you just let the LLMs do your work you won't even find these better ways of doing things. And I'm quite afraid of what the LLMs are going to do to Google, Stackoverflow and documentation in general. In a couple users it'll be ungoogleable too.
  - fzeroracer 2y ago
    As noted LLMs can only give you what you ask for, but for a lot of problems what you ask for isn't what you need; it's two or three steps removed. And LLMs can't tell you that you're doing something wrong; unlike curmudgeonly users on SO or in various forums/channels. My gut feeling is that we're going to enter into a 'dark age' of coding where a lot of previously available resources are going to be ransacked and made hard to find in favor of big corporation owned LLMs. It's already having an extremely bad effect on search in general; we're potentially only a few fights away from sites like SO having users leave en masse. That's why I think having a strong network of engineers to talk with will become more important than ever, almost a return to the IRC days.
  - DrSiemer 2y ago
    Strong disagree. I have learned so many new things since I started using LLM's. Some of them I'm actually embarrassed to admit, because I should have known about them a decade ago. If you work in a small company and you are the most experienced developer, you don't often get feedback on how you can improve things. The trick is, quite simply: just ask. I regularly dump some code I wrote in a language model and then ask what can be done better. I would never do that in any online space, because first, I don't wait an answer maybe some day, I need an answer NOW. And second, I prefer to avoid being called a fool.
anonzzzies 2y ago
Nah, it seems a lot of people are using them wrong? OP seems to also fall in that category. I sit next to many people in pair programming situations and it’s quite weird to see very smart programmers using gpt or copilot; they are the ones shouting ‘stochastic parrot’ on forums and yet expect some kind of mind reading magic when using these tools. When I show that you can put a comment for your function, like you should do anyway, and it writes the function, it usually clicks. Many still find it ‘faster and easier’ to write the function themselves and that’s fine, but, like OP, doing things like entering ???? and generally expecting a useful result will not be optimal.
- ern 2y ago
  You make an interesting point. I once worked with someone who was brilliant, but had gotten into the field via a math degree, and we were building an enterprise-software product. We tried pair-programming, and he cracked within a few minutes. He couldn't articulate his thought processes verbally. Sounds similar to the scenario you're describing.
blibble 2y ago
I wouldn't feel too bad, there's orders more magnitude more money put into this con vs. the last one (NFTs)
- eru 2y ago
  Your counting of 'cons' seems pretty idiosyncratic. You know there have been lots of other technologies people have been working on in the meantime and concurrently? Some of them more questionable, some of them less.
selcuka 2y ago
I basically say the same when people ask me if the AI bubble is going to burst. While I agree that the current push for LLMs should be taken with a grain of salt, I don't think it's a bubble either. I have been finding it useful too, and I don't think I will stop using it because it will look less shiny in the future.
- BlueTemplar 2y ago
  The dot com was a bubble that burst spectacularly enough, yet we still use the World Wide Web, and even some of those websites.
tymscar 2y ago
You are clearly not a dimwit, and you don’t only have way more experience than most people here, you also have some amazing projects under your belt. However I can’t help but notice that the vast majority of blogposts, talks, tweets, and basically everything else you do now is around LLMs. Do you not think that’s indicative of this being “hyped” and “shiny tools”?
- ToValueFunfetti 2y ago
  If hype just means people are excited about it and talking about it, it should be a positive signal about the merit of the thing. I think people get the wrong idea by measuring P(X is good | people are excited about X) and finding it to be low. But P(X is good | people are not excited about X) is vastly lower, and 'hype' as here used is not really discrediting.
  - thaumasiotes 2y ago
    > I think people get the wrong idea by measuring P(X is good | people are excited about X) and finding it to be low. But P(X is good | people are not excited about X) is vastly lower Only if you're including things that don't exist. If you restrict yourself to things that people might conceivably talk about, that probability is actually very high, for the simple reason that bad things are exciting specifically because they are bad. Here are some boring things: - bread - sunlight - air - water - parents Note that starvation, darkness, suffocation, dehydration, and being orphaned are all much more exciting than their opposites.
    - ToValueFunfetti 2y ago
      Poor phrasing on my part; 'good' should have been 'worth talking about'.
- tucnak 2y ago
  Deep learning is the most important thing to happen to philosophy since Socrates / Wittgenstein.
thefaux 2y ago
You are creating a false dichotomy. Very few people are saying there are no benefits, but many have reasonable concerns both about the current efficacy of these tools as well as the expectation of continued exponential growth which is driving much of the current hype. What if we have already passed the inflection point where the exponential growth transitions to an s-curve? That would mean that this technology on its own would only get marginally better than is today. Maybe 2, 4 or even 10x better than it is today, but not 100x or 1000x. To break those barriers, we would need further innovations beyond just throwing more gpus and training data at the problem. I personally am short on LLMs because I believe it is much more likely that we have already crossed the inflection point or will soon. Again, they are impressive but ultimately I think that LLMs will be at best a footnote in history if they are even remembered at all in a few hundred years. But of course I could be wrong.
- simonw 2y ago
  I've been generally staying away from the whole "imagine what this stuff could do next!" side of the discourse. My personal opinion there is that if all research froze today it would still take us years to figure out all of the potential use-cases and applications for the models we have access to right now, and how best to apply them.
vertis 2y ago
No I'm with you. I don't use copilot though for undirected autocomplete, I use tools that allow me to give instructions and diff the results, Cursor and Aider are my current defaults, both with Claude 3 now[0]. Always looking for alternatives though. I think it's always going to be a YMMV experience with LLMs. I'm an extreme generalist with 23 years of experience so it compliments my strengths and weaknesses. I know how to program across 20 odd languages but most of them I need to look stuff up if I'm not using it frequently enough. Now though, don't really need to look stuff up. For me there are two groups, those that want to use LLMs and are pushing it forward and those that are would prefer that LLMs not exist and want it to be hype that goes away. Having followed the hype cycles of VR/AR and crypto I can feel a difference. Both of those felt like a solution in search of a problem. The true believers wanting it to be something like Ready Player One with VR (fun story, Oculus/Facebook actually handed out copies of the novel at Oculus Connect 2 in 2015). The hype around LLMs seems different; like applying a new solution to all the existing problems to see where it helps, and that's just with the current iteration. [0]: I'd prefer to be using local/open equivalents but the capabilities are still lacking.

mhh__ 2y ago

> spam The first 100 lines of any project is spam (usually 0 entropy), so AI is a nice win.

eru 2y ago
That depends on your language, and project.
YurgenJurgensen 2y ago
If the entropy is 0, you didn't need an LLM. If it's the same every time you can just copy and paste those 100 lines from GitHub or StackOverflow.
- mhh__ 2y ago
  Why bother copying and pasting though, that's my whole point. Google is terrible in comparison, just let it write the code for you.

noduerme 2y ago

I think the nomenclature of calling generative AI an "AI" is just hype and that leads to disappointment. After all, it's much more exciting for investors than calling it the world's most expensive textual/audio/visual autocomplete. But that's really what this generatoon currently is. And it can feel like magic, because we're pattern-seeking and pattern-matching creatures, so something that seems to intuit the pattern we're looking for (even if imperfectly) can feel quite a bit like reasoning with another human. To be a bit less generous to humans, it's often the same as talking with another human, because we spend much of our lives writing boilerplate or making small talk or otherwise just kind of on autopilot and answering patterns with expected responses without seriously engaging our complex reasoning. (Do you have to really think about how to map-filter-reduce a dataset anymore? Or think seriously about how you really are when someone says "how are you"? We don't usually expect other people to, and most of the time we're satisfied by exchanging recognizable / somewhat coherent language patterns which we can count on each other to fill in the gaps of. But this is not the intelligence in human intelligence). The corporate hype machine works overtime to hype it as a solution, but that's just what they do. That's all spin. We see through it now, and we're getting to the point where people are asking the only important question about this tech, which is, how much will you pay for a nerfed autocomplete to do X?

tylerchilds 2y ago
“how much will you pay for a needed autocomplete to do X?” i think this is a super important for anyone playing in the LLM space to calculate. currently “AI” providers are selling electricity at a loss to demonstrate product “value”, so even while asking questions is “free” today, there’s actually a finite resource under the hood that needs to have the bill paid in the end. an estimate, but from historic trends, free bills come due in around 7 years
- eru 2y ago
  Even _if_ we don't see our current 'AIs' become more intelligent, I am very certain that we will see the amount of electricity needed to produce 2024-state-of-the-art results drop dramatically over the next few years. People are only just figuring out how any of this works, and they are happy to get any reasonable results at all. They aren't really competing on costs like energy efficiency, yet. --- However I don't really think that 'if' will come to pass: I expect that we will still see lots of advances in the 'intelligence' of these models, so people will compete on these, and worry about electricity usage afterwards.

echelon 2y ago

> shitty clip art AI art is better than most humans, including most artists. You can take the outputs and immediately turn them into animation. Suddenly I, as a single individual, can easily make film content without figuring out set and lighting logistics or roping in dozens of people. This stuff is magic.

ghaff 2y ago
I was somewhat with you through the first sentence then it got a bit breathless. As someone with some graphic design experience but who isn’t very good and can’t really draw, I used Photoshop and it’s genAI capability recently to put together a book cover for an ebook. Would it win any awards? I’m pretty sure not but it’s more than adequate for its purpose and is almost certainly better than anything I could have come up with even if I used some CC artwork. ADDED: I also had something of a vision for what I was looking for and (I think) enough of an eye to fiddle with things and get to the point where I went "This isn't half bad."
- echelon 2y ago
  Exactly! Editing and controllability are getting better almost every week.
YurgenJurgensen 2y ago
I've said it before and I'll say it again: The fact that you didn't think to include an example is telling. If you think AI 'art' is good, show me, and tell me why you think it's good.
- zo1 2y ago
  Not OP. Conflating "art" (the "snobbish" definition if you will) with realistic depictions of imagined worlds, stories and settings is not helpful here. There are so many talented, imaginative people out there that have previously had Zero ability to manifest their thoughts into a visual form. AI has enabled that on so many levels and we should be grateful for it.

015a 2y ago

Yeah; its concerning how depressingly similar Sam's recent language is to, say, SBF's in his prime. Waxing poetically with faux-intellectualism about the sheer addressable market his business will capture (all of humanity) (and funded by the government) [1]. You can just listen to key players in the space, how their language has changed over the past year, and recognize that we're probably very near a local maxima / plateau / AI winter. Reality is: AI is startlingly expensive. Stupid, stupid expensive. Microsoft is building "Stargate", their $100B AI-focused supercomputer [2]. The money to build that isn't coming from selling AI services. Its coming from their old, boring, tried and true businesses; Windows, Office, Azure, and M365. Xbox is dying before our eyes. Their tried and true businesses make great money but aren't growing. Its up to AI to prove to investors that their P/E multiple is justified. [1] https://twitter.com/tsarnick/status/1789107043825262706 https://twitter.com/tsarnick/status/1789107043825262706 [2] https://www.tomsguide.com/ai/meet-stargate-the-dollar100-billion-ai-supercomputer-being-built-by-microsoft-and-openai https://www.tomsguide.com/ai/meet-stargate-the-dollar100-bil...

viking123 2y ago
I always mix SBF and Altman in my mind, and have to spend few seconds thinking which was which.. But IMO the current AI hype has some almost religious elements to it

dogcomplex 2y ago

It is honestly shocking how different people can be, even within programming. Been finding LLMs very useful for pair programming and low-effort spin-ups of projects I would just never do without the tools. Frankly, I'm waiting any day for an architecture to arise that takes lossy LLMs and has them error correct enough to produce 99.9999% reliable progress on simple tasks cheaply. This feels incredibly close, and if the unit of intelligence of each step of those is around GPT4 level? Jesus christ. Programming massively automated overnight. Any programmer that is this dismissive of these tools frankly... hasn't been using them right, and has no imagination - or a massive ego. This stuff is still far too early to make such judgements.

dartos 2y ago

I don’t really think AI will get there the way we’re going about it. Statistical models will always be limited by their data and how edge case filled reality is.

noobermin 2y ago

People are either misusing this meme or it makes no sense.

CaptainFever 2y ago
It's likely an ironic usage.

mondrian 2y ago

It's definitely useful, but in higher level chat use cases, not so much high-precision generation (yet). Rubber ducking, brainstorming, and search-like use cases are definitely a level above what we had.

n_ary 2y ago

> It's a pretty clear depiction of the investor class being so detached from the technical reality I think, their ROI model is much different and subtle. Few decades back, everything ran on-premises and costed ok ... Then a huge investment had been shovelled into fancy frameworks and pushing strange paradigms (microservices, distributed systems, monoliths whatever flavour works for the tech giants) so now people just put together whatever and needs massive _cloud_ to run basic things that used to run on few MHz devices. In the end, the ROI is NOT from these models, those are just gambles in case one actually becomes a king maker. The actual ROI is from the _cloud_ where innocent people will rent expensive hardware to try and utilise these models and funneling wealth to the actual investments(cloud operators, hardware vendors) which are already part of portfolio of these investors. Basically, $1MM invested in random shiny startup will inspire other shiny startups to also spend similar in race to become unicorn, all the while the cloud operators and GPU vendors are laughing to the bank preparing their shareholders(usually same investors) dividends.