15 ms·

>But superhuman AI seems now only few years away Seems unreasonable. You are afraid because marketing gurus like Altman made you believe that a frog that can m

>But superhuman AI seems now only few years away Seems unreasonable. You are afraid because marketing gurus like Altman made you believe that a frog that can make bigger leap than before will be able to fly.

cubefox 2y ago
No, because we have seen massive improvements in AI over the last years, and all the evidence points to this progress continuing at a fast pace.
- StrLght 2y ago
  Extrapolation of past progress isn't evidence.
  - cubefox 2y ago
    Past progress is evidence for future progress.
    - moe_sc 2y ago
      Might be an indicator, but it isn't evidence.
    - StrLght 2y ago
      That's probably what every self-driving car company thought ~10 years ago or so, everything was moving so fast for them back then. Now it doesn't seem like we're getting close to solution for this. Surely this time it's going to be different, AGI is just around a corner. /s
      - johnthewise 2y ago
        Would you have predicted in summer of 2022 that gpt4 level conversational agent is a possibility in the next 5 years? People have tried to do it in the past 60 years and failed. How is this time not different? On a side note, I find this type of critique of what future of tech might look like the most uninteresting one. Since tech by nature inspiries people about the future, all tech get hyped up. all you gotta do then is pick any tech, point out people have been wrong, and ask how likely is it that this time it is different.
        StrLght 2y ago
        Unfortunately, I don't see any relevance in that argument, if you consider GPT-4 to be a breakthrough -- then sure, single breakthroughs happen, I am not arguing with that. Actually, same thing happened with self-driving: I don't think many people expected Tesla to drop FSD publicly back then. Now, chain of breakthroughs happening in a small timeframe? Good luck with that.
        cubefox 2y ago
        We have seen multiple massive AI breakthroughs in the last few years.
        Jensson 2y ago
        They are the same breakthrough applied to different domains, I don't see them as different. We will need a new breakthrough, not applying the same solution to new things.
        StrLght 2y ago
        Which ones are you referring to? Just to make it clear, I see only 1 breakthrough [0]. Everything that happened afterwards is just application of this breakthrough with different training sets / to different domains / etc. [0]: https://en.wikipedia.org/wiki/Attention_Is_All_You_Need https://en.wikipedia.org/wiki/Attention_Is_All_You_Need
        cubefox 2y ago
        Autoregressive language models, the discovery of the Chinchilla scaling law, MoEs, supervised fine-tuning, RLHF, whatever was used to create OpenAI o1, diffusion models, AlphaGo, AlphaFold, AlphaGeometry, AlphaProof.
      - mitthrowaway2 2y ago
        If you wake up from a coma and see the headline "Today Waymo has rolled out a nationwide robotaxi service", what year do you infer that it is?
    - nitwit005 2y ago
      Not exactly. If you focus in on a single technology, you tend to see rapid improvement, followed by slower progress. Sometimes this is masked by people spending more due to the industry becoming more important, but it tends to be obvious over the longer term.
  - mitthrowaway2 2y ago
    You don't have to extrapolate. There's a frenzy of talent being applied to this problem, it's drawing more brainpower the more progress that is made. Young people see this as one of the most interesting, prestigious, and best-paying fields to work in. A lot of these researchers are really talented, and are doing more than just scaling up. They're pushing at the frontiers in every direction, and finding methods that work. The progress is broadening; it's not just LLMs, it's diffusion models, it's SLAM, it's computer vision, it's inverse problems, it's locomotion. The tooling is constantly improving and being shared, lowering the barrier to entry. And classic "hard problems" are yielding in the process. It's getting hard to even find hard problems any more. I'm not saying this as someone cheering this on; I'm alarmed by it. But I can't pretend that it's running out of steam. It's possible it will run out of money, but even if so, only for a while.
    - leptons 2y ago
      The AI bubble is already starting to burst. They Sam Altmans' of the world over-sold their product and over-played their hand by suggesting AGI is coming. It's not. What they have is far, far, far from AGI. "AI" is not going to be as important as you think it is in the near future, it's just the current tech-buzz and there will be something else that takes its place, just like when "web 2.0" was the new hotness.
      - kranuck 2y ago
        It's gonna be massive because companies love to replace humans at any opportunity and they don't care at all about quality in a lot of places. For example, why hire any call center workers? They already outsourced the jobs to the lowest bidder and their customers absolutely hate it. Fire those people and get some AI in there so it can provide shitty service for even cheaper. In other words, it will just make things a bit worse for everyone but those at the very top. usual shit.
    - mvdtnz 2y ago
      > There's a frenzy of talent being applied to this problem, it's drawing more brainpower the more progress that is made. Young people see this as one of the most interesting, prestigious, and best-paying fields to work in. A lot of these researchers are really talented, and are doing more than just scaling up. They're pushing at the frontiers in every direction, and finding methods that work. You could have seen this exact kind of thing written 5 years ago in a thread about blockchains.
  - coryfklein 2y ago
    Do you expect the hockeystick graph of technological development since the industrial evolution to slow? Or that it will proceed, only without significant advances in AI? Seems like the base case here is for the exponential growth to continue, and you'd need a convincing argument to say otherwise.
    - StrLght 2y ago
      Which chart are you referencing exactly? How does it define technological development? It's nearly impossible for me to discuss a chart without knowing what axis refer. Without specifics all I can say is that I don't acknowledge any measurable benefits of AI (in its' current state) in real world applications. So I'd say I am leaning towards latter.
    - kranuck 2y ago
      That's no guarantee that AI continues advancing at the same pace, and no one has been arguing against overall technological progress slowing Refining technology is easier than the original breakthrough, but it doesn't usually lead to a great leap forward. LLMs were the result of breakthroughs, but refining them isn't guaranteed to lead to AGI. It's not guaranteed (or likely) to improve at an exponential rate.
- lawn 2y ago
  The evidence I've been seeing is that progress with LLMs have already slowed down and that they're nowhere near good enough to replace programmers. They can be useful tools ro be sure, but it seems more and more clear that they will not reach AGI.
  - cubefox 2y ago
    They are already above average human level on many tasks, like math benchmarks.
    - kranuck 2y ago
      If you ignore the part where there proofs are meandering drivel, sure.
      - cubefox 2y ago
        Even if you don't ignore this part they (e.g. o1-preview) are still better at proofs than the average human. Substantially better even.
    - cudgy 2y ago
      So are calculators …
    - lawn 2y ago
      Yes, there are certain tasks they're great at, just as AI has been superhuman in some tasks for decades.
      - cubefox 2y ago
        But now they are good or even great at way more tasks than before because they can understand and use natural languages like English.
        lawn 2y ago
        Yeah, and they're still under delivering to their hype and the improvements have vastly slowed down.
- mvdtnz 2y ago
  Does it though? I have seen the progress basically stop at "shitty sentence generator that can't stop lying".
- Hercuros 2y ago
  I think the biggest fallacy in this type of thinking is that it projects all AI progress into a single quantity of “intelligence” and then proceeds to extrapolate that singular quantity into some imagined absurd level of “superintelligence”. In reality, AI progress and capabilities are not so reducible to singular quantities. For example, it’s not clear that we will ever get rid of the model’s tendencies to just produce garbage or nonsense sometimes. It’s entirely possible that we remain stuck at more incremental improvements now, and I think the bogeyman of “superintelligence” needs to be much more clearly defined rather than by extrapolation of some imagined quantity. Or maybe we reach a somewhat human-like level, but not this imagined “extra” level of superintelligence. Basically the argument is something to the effect of “big will become bigger and bigger, and then it will become like SUPER big and destroy us all”.
- rocho 2y ago
  But that does not prove anything. We don't know where we are on the AI-power scale currently. "Superintelligence", whatever that means, could be 1 year or 1000 years away at our current progress, and we wouldn't know until we reach it.
  - handoflixue 2y ago
    50 years ago we could rather confidently say that "Superintelligence" was absolutely not happening next year, and was realistically decades ago. If we can say "it could be next year", then things have changed radically and we're clearly a lot closer - even if we still don't know how far we have to go. A thousand years ago we hadn't invented electricity, democracy, or science. I really don't think we're a thousand years away from AI. If intelligence is really that hard to build, I'd take it as proof that someone else must have created us humans.
    - 110 2y ago
      Umm, customary, tongue-in-cheek reference to McCarthy's proposal for a 10 person research team to solve AI in 2 months (over the Summers)[1]. This was ~70 years ago :) Not saying we're in necessarily the same situation. But it remains difficult to evaluate effort required for actual progress. [1]: https://www-formal.stanford.edu/jmc/history/dartmouth/dartmouth.html https://www-formal.stanford.edu/jmc/history/dartmouth/dartmo...
khafra 2y ago
> If an elderly but distinguished scientist says that something is possible, he is almost certainly right - Arthur C. Clarke Geoffrey Hinton is a 76 year old Turing Award* winner. What more do you want? *Corrected by kranner
- kranner 2y ago
  > Geoffrey Hinton is a 76 year old Nobel Prize winner. Turing Award, not Nobel Prize
  - khafra 2y ago
    Thanks for the correction; I am undistinguished and getting more elderly by the minute.
  - khafra 2y ago
    Reality has now corrected my error, which was amongst the funniest possible outcomes.
    - kranner 2y ago
      Indeed! Your comment was the first thing I thought of when I heard the news and I thought of replying too but assumed you might not have enabled notifications Hilarious, all in all!
- nessbot 2y ago
  This is like a second-order appeal to authority fallacy, which is kinda funny.
- randomdata 2y ago
  Hinton says that superintelligence is still 20 years away, and even then he only gives his prediction a 50% chance. A far cry from the few year claim. You must be doing that "strawberry" thing again? To us humans, A-l-t-m-a-n is not H-i-n-t-o-n.
  - 2y ago
AI_beffr 2y ago
wrong. i was extremely concerned in 2018 and left many comments almost identical to this one back then. this was based off of the first gtp samples that openai released to the public. there was no hype or guru bs back then. i believed it because it was obvious. it was obvious then and it is still obvious today.
klabb3 2y ago
Plus it’s not even defined what superhuman AI means. A calculator sure looked superhuman when it was invented. And it is! Another analogy is breeding and racial biology which used to be all the hype (including in academia). The fact that humans could create dogs from wolves, looked almost limitless with the right (wrong) glasses. What we didn’t know is that wolf had a ton of genes that played a magic trick where a diversity we couldn’t perceive was there all along, in the genetic material, and it we just helped make it visible. Ie a game of diminishing returns. Concretely for AI, it has shown us that pattern matching and generation are closely related (well I have a feeling this wasn’t surprising to neuro-scientists). And also that they’re more or less domain agnostic. However, we don’t know whether pattern matching alone is “sufficient”, and if not, what exactly and how hard “the rest” is. Ai to me feels like a person who had a stroke, concussion or some severe brain injury, it can appear impressively able in a local context, but they forgot their name and how they got there. They’re just absent.
8338550bff96 2y ago
Flying is a good analogy. Superman couldn't fly, but at some point when you can jump so far there isn't much of a difference
- latexr 2y ago
  There is an enormous difference. Flying allows you to stop, change direction, make corrections, and target with a large degree of accuracy. Jumping leaves you at the mercy of your initial calculations. If you jumped in a way that you’ll land inside a volcano, all you can do in your last moments is watch and wait for your demise.
  - 8338550bff96 2y ago
    A volcano can't kill superman. Rebuttal rejected
digging 2y ago
That argument holds no water because the grifters aren't the source of this idea. I literally don't believe Altman at all; his public words don't inspire me to agree or disagree with them - just ignore them. But I also hold the view that transformative AI could be very close. Because that's what many AI experts are also talking about from a variety of angles. Additionally, when you're talking with certainty about whether transformative AI is a few years away or not, that's the only way to be wrong. Nobody is or can be certain, we can only have estimations of various confidence levels. So when you say "Seems unreasonable", that's being unreasonable.
- kranuck 2y ago
  > Because that's what many AI experts are also talking about from a variety of angles. Wow, in that case I'm convinced. Such an unbiased group with nothing at all to gain from massive AI hype.

cubefox 2y ago

No, because we have seen massive improvements in AI over the last years, and all the evidence points to this progress continuing at a fast pace.

StrLght 2y ago
Extrapolation of past progress isn't evidence.
- cubefox 2y ago
  Past progress is evidence for future progress.
  - moe_sc 2y ago
    Might be an indicator, but it isn't evidence.
  - StrLght 2y ago
    That's probably what every self-driving car company thought ~10 years ago or so, everything was moving so fast for them back then. Now it doesn't seem like we're getting close to solution for this. Surely this time it's going to be different, AGI is just around a corner. /s
    - johnthewise 2y ago
      Would you have predicted in summer of 2022 that gpt4 level conversational agent is a possibility in the next 5 years? People have tried to do it in the past 60 years and failed. How is this time not different? On a side note, I find this type of critique of what future of tech might look like the most uninteresting one. Since tech by nature inspiries people about the future, all tech get hyped up. all you gotta do then is pick any tech, point out people have been wrong, and ask how likely is it that this time it is different.
      - StrLght 2y ago
        Unfortunately, I don't see any relevance in that argument, if you consider GPT-4 to be a breakthrough -- then sure, single breakthroughs happen, I am not arguing with that. Actually, same thing happened with self-driving: I don't think many people expected Tesla to drop FSD publicly back then. Now, chain of breakthroughs happening in a small timeframe? Good luck with that.
        cubefox 2y ago
        We have seen multiple massive AI breakthroughs in the last few years.
        Jensson 2y ago
        They are the same breakthrough applied to different domains, I don't see them as different. We will need a new breakthrough, not applying the same solution to new things.
        StrLght 2y ago
        Which ones are you referring to? Just to make it clear, I see only 1 breakthrough [0]. Everything that happened afterwards is just application of this breakthrough with different training sets / to different domains / etc. [0]: https://en.wikipedia.org/wiki/Attention_Is_All_You_Need https://en.wikipedia.org/wiki/Attention_Is_All_You_Need
        cubefox 2y ago
        Autoregressive language models, the discovery of the Chinchilla scaling law, MoEs, supervised fine-tuning, RLHF, whatever was used to create OpenAI o1, diffusion models, AlphaGo, AlphaFold, AlphaGeometry, AlphaProof.
    - mitthrowaway2 2y ago
      If you wake up from a coma and see the headline "Today Waymo has rolled out a nationwide robotaxi service", what year do you infer that it is?
  - nitwit005 2y ago
    Not exactly. If you focus in on a single technology, you tend to see rapid improvement, followed by slower progress. Sometimes this is masked by people spending more due to the industry becoming more important, but it tends to be obvious over the longer term.
- mitthrowaway2 2y ago
  You don't have to extrapolate. There's a frenzy of talent being applied to this problem, it's drawing more brainpower the more progress that is made. Young people see this as one of the most interesting, prestigious, and best-paying fields to work in. A lot of these researchers are really talented, and are doing more than just scaling up. They're pushing at the frontiers in every direction, and finding methods that work. The progress is broadening; it's not just LLMs, it's diffusion models, it's SLAM, it's computer vision, it's inverse problems, it's locomotion. The tooling is constantly improving and being shared, lowering the barrier to entry. And classic "hard problems" are yielding in the process. It's getting hard to even find hard problems any more. I'm not saying this as someone cheering this on; I'm alarmed by it. But I can't pretend that it's running out of steam. It's possible it will run out of money, but even if so, only for a while.
  - leptons 2y ago
    The AI bubble is already starting to burst. They Sam Altmans' of the world over-sold their product and over-played their hand by suggesting AGI is coming. It's not. What they have is far, far, far from AGI. "AI" is not going to be as important as you think it is in the near future, it's just the current tech-buzz and there will be something else that takes its place, just like when "web 2.0" was the new hotness.
    - kranuck 2y ago
      It's gonna be massive because companies love to replace humans at any opportunity and they don't care at all about quality in a lot of places. For example, why hire any call center workers? They already outsourced the jobs to the lowest bidder and their customers absolutely hate it. Fire those people and get some AI in there so it can provide shitty service for even cheaper. In other words, it will just make things a bit worse for everyone but those at the very top. usual shit.
  - mvdtnz 2y ago
    > There's a frenzy of talent being applied to this problem, it's drawing more brainpower the more progress that is made. Young people see this as one of the most interesting, prestigious, and best-paying fields to work in. A lot of these researchers are really talented, and are doing more than just scaling up. They're pushing at the frontiers in every direction, and finding methods that work. You could have seen this exact kind of thing written 5 years ago in a thread about blockchains.
- coryfklein 2y ago
  Do you expect the hockeystick graph of technological development since the industrial evolution to slow? Or that it will proceed, only without significant advances in AI? Seems like the base case here is for the exponential growth to continue, and you'd need a convincing argument to say otherwise.
  - StrLght 2y ago
    Which chart are you referencing exactly? How does it define technological development? It's nearly impossible for me to discuss a chart without knowing what axis refer. Without specifics all I can say is that I don't acknowledge any measurable benefits of AI (in its' current state) in real world applications. So I'd say I am leaning towards latter.
  - kranuck 2y ago
    That's no guarantee that AI continues advancing at the same pace, and no one has been arguing against overall technological progress slowing Refining technology is easier than the original breakthrough, but it doesn't usually lead to a great leap forward. LLMs were the result of breakthroughs, but refining them isn't guaranteed to lead to AGI. It's not guaranteed (or likely) to improve at an exponential rate.
lawn 2y ago
The evidence I've been seeing is that progress with LLMs have already slowed down and that they're nowhere near good enough to replace programmers. They can be useful tools ro be sure, but it seems more and more clear that they will not reach AGI.
- cubefox 2y ago
  They are already above average human level on many tasks, like math benchmarks.
  - kranuck 2y ago
    If you ignore the part where there proofs are meandering drivel, sure.
    - cubefox 2y ago
      Even if you don't ignore this part they (e.g. o1-preview) are still better at proofs than the average human. Substantially better even.
  - cudgy 2y ago
    So are calculators …
  - lawn 2y ago
    Yes, there are certain tasks they're great at, just as AI has been superhuman in some tasks for decades.
    - cubefox 2y ago
      But now they are good or even great at way more tasks than before because they can understand and use natural languages like English.
      - lawn 2y ago
        Yeah, and they're still under delivering to their hype and the improvements have vastly slowed down.
mvdtnz 2y ago
Does it though? I have seen the progress basically stop at "shitty sentence generator that can't stop lying".
Hercuros 2y ago
I think the biggest fallacy in this type of thinking is that it projects all AI progress into a single quantity of “intelligence” and then proceeds to extrapolate that singular quantity into some imagined absurd level of “superintelligence”. In reality, AI progress and capabilities are not so reducible to singular quantities. For example, it’s not clear that we will ever get rid of the model’s tendencies to just produce garbage or nonsense sometimes. It’s entirely possible that we remain stuck at more incremental improvements now, and I think the bogeyman of “superintelligence” needs to be much more clearly defined rather than by extrapolation of some imagined quantity. Or maybe we reach a somewhat human-like level, but not this imagined “extra” level of superintelligence. Basically the argument is something to the effect of “big will become bigger and bigger, and then it will become like SUPER big and destroy us all”.
rocho 2y ago
But that does not prove anything. We don't know where we are on the AI-power scale currently. "Superintelligence", whatever that means, could be 1 year or 1000 years away at our current progress, and we wouldn't know until we reach it.
- handoflixue 2y ago
  50 years ago we could rather confidently say that "Superintelligence" was absolutely not happening next year, and was realistically decades ago. If we can say "it could be next year", then things have changed radically and we're clearly a lot closer - even if we still don't know how far we have to go. A thousand years ago we hadn't invented electricity, democracy, or science. I really don't think we're a thousand years away from AI. If intelligence is really that hard to build, I'd take it as proof that someone else must have created us humans.
  - 110 2y ago
    Umm, customary, tongue-in-cheek reference to McCarthy's proposal for a 10 person research team to solve AI in 2 months (over the Summers)[1]. This was ~70 years ago :) Not saying we're in necessarily the same situation. But it remains difficult to evaluate effort required for actual progress. [1]: https://www-formal.stanford.edu/jmc/history/dartmouth/dartmouth.html https://www-formal.stanford.edu/jmc/history/dartmouth/dartmo...

khafra 2y ago

> If an elderly but distinguished scientist says that something is possible, he is almost certainly right - Arthur C. Clarke Geoffrey Hinton is a 76 year old Turing Award* winner. What more do you want? *Corrected by kranner

kranner 2y ago
> Geoffrey Hinton is a 76 year old Nobel Prize winner. Turing Award, not Nobel Prize
- khafra 2y ago
  Thanks for the correction; I am undistinguished and getting more elderly by the minute.
- khafra 2y ago
  Reality has now corrected my error, which was amongst the funniest possible outcomes.
  - kranner 2y ago
    Indeed! Your comment was the first thing I thought of when I heard the news and I thought of replying too but assumed you might not have enabled notifications Hilarious, all in all!
nessbot 2y ago
This is like a second-order appeal to authority fallacy, which is kinda funny.
randomdata 2y ago
Hinton says that superintelligence is still 20 years away, and even then he only gives his prediction a 50% chance. A far cry from the few year claim. You must be doing that "strawberry" thing again? To us humans, A-l-t-m-a-n is not H-i-n-t-o-n.
- 2y ago

AI_beffr 2y ago

wrong. i was extremely concerned in 2018 and left many comments almost identical to this one back then. this was based off of the first gtp samples that openai released to the public. there was no hype or guru bs back then. i believed it because it was obvious. it was obvious then and it is still obvious today.

klabb3 2y ago

Plus it’s not even defined what superhuman AI means. A calculator sure looked superhuman when it was invented. And it is! Another analogy is breeding and racial biology which used to be all the hype (including in academia). The fact that humans could create dogs from wolves, looked almost limitless with the right (wrong) glasses. What we didn’t know is that wolf had a ton of genes that played a magic trick where a diversity we couldn’t perceive was there all along, in the genetic material, and it we just helped make it visible. Ie a game of diminishing returns. Concretely for AI, it has shown us that pattern matching and generation are closely related (well I have a feeling this wasn’t surprising to neuro-scientists). And also that they’re more or less domain agnostic. However, we don’t know whether pattern matching alone is “sufficient”, and if not, what exactly and how hard “the rest” is. Ai to me feels like a person who had a stroke, concussion or some severe brain injury, it can appear impressively able in a local context, but they forgot their name and how they got there. They’re just absent.

8338550bff96 2y ago

Flying is a good analogy. Superman couldn't fly, but at some point when you can jump so far there isn't much of a difference

latexr 2y ago
There is an enormous difference. Flying allows you to stop, change direction, make corrections, and target with a large degree of accuracy. Jumping leaves you at the mercy of your initial calculations. If you jumped in a way that you’ll land inside a volcano, all you can do in your last moments is watch and wait for your demise.
- 8338550bff96 2y ago
  A volcano can't kill superman. Rebuttal rejected

digging 2y ago

That argument holds no water because the grifters aren't the source of this idea. I literally don't believe Altman at all; his public words don't inspire me to agree or disagree with them - just ignore them. But I also hold the view that transformative AI could be very close. Because that's what many AI experts are also talking about from a variety of angles. Additionally, when you're talking with certainty about whether transformative AI is a few years away or not, that's the only way to be wrong. Nobody is or can be certain, we can only have estimations of various confidence levels. So when you say "Seems unreasonable", that's being unreasonable.

kranuck 2y ago
> Because that's what many AI experts are also talking about from a variety of angles. Wow, in that case I'm convinced. Such an unbiased group with nothing at all to gain from massive AI hype.