23 ms·

> This entire piece is based on one massive, unsupported assertion, which is that LLM progress will cease. Which is countered by...the assertion that it won't?

> This entire piece is based on one massive, unsupported assertion, which is that LLM progress will cease. Which is countered by...the assertion that it won't? LLMs won't get intelligent. That's a fact based on their MO. They are sequence completion engines. They can be fine tuned to specific tasks, but at their core, they remain stochastic parrots. > I want to know only one thing, which is what gives him the confidence necessary to say that. I want to know only one thing, what gives the confidence to say otherwise?

rhn_mk1 3y ago
It's countered by not making the assertion and not being able to make conclusions. You only need a lack of confidence for that. It's orders of magnitude easier to not have knowledge compared to having it.
beefield 3y ago
> LLMs won't get intelligent. That's a fact based on their MO. I kind of agree. However, I see a real possibility that in the near future LLM behaviour would be practically indistinguishable from intelligent/sentient behavior. And at that point we (or at least I) are facing some really interesting/difficult questions, namely how do you know an intelligent looking thing actually is intelligent (or sentient). How do you prove me you/LLM are/aren't a philosophical zombie? How we are supposed to treat very much intelligent/sentient looking things when we are not sure if they are sentient/intelligent or not? Let's face it, lots of people are dumb as rock (too often very much me included). Why we should be able to treat something badly just because we think we know they can't be intelligent, even if they walk , look and quack like intelligent duck? I personally have started to think that the behavior of humans should be judged by the behaviour, not the target. If you want to behave like an asshole towards a teddy bear, then you most likely are an asshole.
- savolai 3y ago
  It seems to me that the real question here is what is true human intelligence. Ai has made it plain to see, by being able to replicate it so convincingly, that much of what we have considered intelligence has been pattern matching or acting as complex parrots. There is much more to the abilities of human body-mind-emotional-experiential being, but it is only slowly becoming mainstream. (Edit: Of course there are also many analytical skills that AI cannot match at this point. My point is that we shouldn’t overlook any area of human capacity.) One salient question in this is: will we reach a level of intelligence where we become beings capable of actual collaboration that doesn’t waste so much effort in conflicts, or one that is capable of living in harmony within its environment? What capabilities of awareness, trauma work, emotional maturity and self reflection does this require? What resources hidden inside humanity that we have forgotten do we need to wield? Does AI have something to contribute to this process happening?
  - worrycue 3y ago
    > It seems to me that the real question here is what is true human intelligence. IMHO the main weakness with LLMs is they can’t really reason. They can statistically guess their way to an answer - and they do so surprisingly well I will have to admit - but they can’t really “check” themselves to ensure what they are outputting makes any sense like humans do (most of the time) - hence the hallucinations.
    - NumberWangMan 3y ago
      Apparently GPT-4 is getting pretty good at knowing when it's wrong: https://thezvi.substack.com/p/ai-26-fine-tuning-time#%C2%A7gpt-real-this-time https://thezvi.substack.com/p/ai-26-fine-tuning-time#%C2%A7g... (They asked GPT-3.5 and GPT-4 "are you sure" to see if it would change its answer, both when the original answer was right, and when it was wrong)
      - worrycue 3y ago
        Does it do that because it can check it’s own reasoning? Or is it just doing so because OpenAI programmed it to not show alternative answers if the probability of the current answer being right is significantly higher than the alternatives?
        NumberWangMan 3y ago
        I don't know. I don't think anyone is directly programming GPT-4 to behave in any way, they're just training it to give the responses they want, and it learns. Something inside it seems to be figuring out some way of representing confidence in its own answers, and reacting in the appropriate way, or perhaps it is checking its own reasoning. I don't think anyone really knows at this point.
    - naasking 3y ago
      As the other poster said, they can check themselves but this requires an iterative process where the output is fed back in as input. Think of LLMs as the output of a human's stream of consciousness: it is intelligent, but has a high chance of being riddled with errors. That's why we iterate on our first thoughts to refine them.
- velvetz 3y ago
  Off-topic and might sound strange but I find it intriguing that you wrote behavior and behaviour in these two different forms in the same sentence:)
  - coldtea 3y ago
    Quite easy to happen if like many of us you were taugh British English, and then you remember that you're on a US forum, and everybody around you uses the US spelling for things (and you get to read the US variants all the time in other comments).
  - beefield 3y ago
    I don't have any excuses but being non native in english having exposure to both British and US english seems to confuse my brain.
- js8 3y ago
  Any sufficiently advanced intelligence is indistinguishable from an LLM? Cute, but practically speaking, I would prefer the former.
albertzeyer 3y ago
Look at the scaling laws. We found that extrapolating the performance given a few data points with smaller models is actually very accurate. That's how they determined hyper parameters, by tuning them on multiple smaller scale models and then extrapolating. So far, all those predictions were quite good. Together with a bigger model, we also need more data to get better performance. If we add video and audio to the text data, we have still a lot more data we can use, so this is also not really a problem. It would be very unexpected that those scaling laws are suddenly not true anymore for the next order of magnitude in model and data size.
- dontupvoteme 3y ago
  Scaling laws apply to a single model. The best single model right now is supposedly a 8x mixture of experts, so not even really a single model in the purist sense. I still expect the final solution will be more along the lines of picking the best model(s) from a sea of possible models, switching them in and out as needed, and then automatically reiterating as needed.
golol 3y ago
>LLMs won't get intelligent. That's a fact based on their MO. They are sequence completion engines. They can be fine tuned to specific tasks, but at their core, they remain stochastic parrots. This is absolutely wrong. There is nothing about their MO that stops them from being intelligent. Suppose I build a human LLM as follows: A random human expert is picked and he is shown the current context window. He is given 1 week to deliberate and then may choose the next word/token/character. Then you hook this human LLM into an auto-GPT style loop. There is no reason it couldn't operate with high intelligence on text data. Not also that LLMs are not really about language at all anymore, the architectures can be used on any sequence data. Right now we are compute limited. If compute was 100x cheaper we could have GPT-6, bring 100x bigger, we could have really large and complex agents using GPT-4 power models, or we could train on tupled text-video data of subtitles videos. Given the world model LLMs manage to learn out of text data, I am 100% certain that a sufficiently large transformer can learn a decent world model from text-video data. Then our agents could also have a good physical understanding.
- ambrozk 3y ago
  Humans will never be intelligent. They're optimized for producing offspring, not reasoning. Humans may appear to be intelligent from a distance, but talk to one for any length of time and you'll find they make basic errors of reasoning that no truly thinking being would fall for. /s
  - ksaho 3y ago
    Take out the /s tag and you are right on the money. Humans can not be trusted with anything because they are trivially fallible. Humans are terribly stupid, destroy their own societies and refuse to see reason. They also hallucinate when their destructive tendencies start catching up to them.
    - ambrozk 3y ago
      If the most intelligent machines ever observed in the universe do not count as "intelligent," then we have a semantic, and not a substantive difference of opinion.
squeaky-clean 3y ago
> LLMs won't get intelligent. I think this sentence doesn't mean much unless we have a strict definition of what intelligence means. Just today ChatGPT helped me solve a DNS issue that I would not have been able to solve on my own in one day, let alone an hour. I'd consider it already more intelligent than myself when it comes to DNS.
- Aperocky 3y ago
  It's seen more DNS content than you and anybody else have seen in their entire lives, and are able to regurgitate what it read because it has far faster memory access than you did. A dictionary contain knowledge but no intelligence.
  - TeMPOraL 3y ago
    And LLMs are the opposite of dictionary, actually. They suck at storing facts. They excel at extracting patterns from noise and learning them. It's not obvious to me that this isn't intelligence; on the contrary, I feel it's very much a core component of it.
    - BartjeD 3y ago
      Isn't that a dictionary of patterns?
    - worrycue 3y ago
      > They excel at extracting patterns from noise and learning them. You can argue those are facts too.
      - lucubratory 3y ago
        Yes, I have noticed that a lot of extreme AI cynics have been arguing that any and every example of reasoning or thinking that an LLM displays is just some variant of memorisation.
        3y ago
edf13 3y ago
> They are sequence completion engines But at the basic level - isn't our own brain just a sequence completion engine too?
- HPsquared 3y ago
  The brain does seem to do a lot of pattern-matching and prediction.
nopinsight 3y ago
Geoffrey Hinton, Andrew Ng, and quite a few other top AI researchers believe that current LLMs (and incoming waves of multimodal LFMs) learn world models; they are not simply 'stochastic parrots'. If one feeds GPT-4 a novel problem that does not require multi-step reasoning or very high precision to solve, it can often solve it.
- auggierose 3y ago
  Anyone who has worked a bit with a top LLM thinks that they learn world models. Otherwise, what they are doing would be impossible. I've used them for things that are definitely not on the web, because they are brand new research. They are definitely able to apply what they've learnt in novel ways.
  - RandomLensman 3y ago
    If they learn world models, those world models are incredible poor, i.e., there is no consistency of thought in those world models. In my experience, things outside coding quickly devolve into something more like "technobabble" (and in coding there is always a lot of made-up stuff that doesn't exists in terms of functions etc.).
    - coldtea 3y ago
      >If they learn world models, those world models are incredible poor, i.e., there is no consistency of thought in those world models Incredibly poor compared to ours, but thousands of times better than what "AI" we had before.
      - RandomLensman 3y ago
        Not sure that matters much as they are only for low risk stuff without skilled supervision, so back to advertising, marketing, cheap customer support, etc.
    - auggierose 3y ago
      I see them more as creative artists who have very good intuition, but are poor logicians. Their world model is not a strict database of consistent facts, it is more like a set of various beliefs, and of course those can be highly contradictory.
roenxi 3y ago
I'm put in mind of the OpenAI DoTA bot that was winning 99% of its games and some people refused to admit that it knew how to play DoTA based on some esoteric interpretation of the word "play". We're going to see exponential increases in processing power of the best GPU clusters and human brains are a stationary target. And there is precious little evidence that the average human is much more than an LLM. LLMs are already more likely to understand a topic to a high standard than a given human. They're going to progress and if they aren't intelligent then intelligence is overrated and I'd rather have whatever they have.
alpaca128 3y ago
> LLMs won't get intelligent Even assuming that is true: LLMs aren't all that exists in AI research and just like LLMs are amazing in terms of language it's possible similar breakthroughs could be made in more abstracted areas that could use LLMs for IO. If you think ChatGPT is nice, wait for ChatGPT as frontend for another AI that doesn't have to spend a single CPU cycle on language.
- oceanplexian 3y ago
  The next AI wave hasn't even started. Imagine an LLM the size of GPT-4 but it's trained on nothing but gene sequence completion. All the models being used in academia are basically toys, none of those guys are running hardware at a scale that can even remotely touch Azure, Meta, etc, and right now there is a massive global shortage of GPU compute that's eventually going to clear up. We know models get A LOT better when they are scaled up and are fed more data, so why wouldn't the same be true for other problems besides text completion?
  - td118 3y ago
    [dead]
- benterix 3y ago
  > LLMs aren't all that exists in AI research Frankly, I'm a bit worried about all the rest now that LLMs proved to be so successful. We might exploit them and arrive to a dead end. In the meantime, other potentially crucial developments in AI might get less attention and funding.
glenstein 3y ago
>Which is countered by...the assertion that it won't? No it's countered by principled restraint in not making an affirmative claim one way or the other. I've heard this referred to as the overconfident pessimism problem. Which is that normal, well founded scientific discipline and evidence-based restraint go out of the window when people declare, without evidence that they know certain advances won't happen. Because people get mentally trapped into this framing of either have to declare that it will happen or that it won't, seeming to forget that you can just adopt the position of modesty and say the dust hasn't yet settled.
- tlarkworthy 3y ago
  AI has been around the corner since the 1950s, this is the historical evidence for the pessimistic stance against over optimistic predictions. LLMs are a huge stride forward, but AI does not progress like Moore's law. LLM have revealed a new wall. Combining multi agents is not working out as hoped.
  - glenstein 3y ago
    Perhaps without intending to, you've cited a pretty appropriate example of overconfident pessimism. Philosopher Hubert Dreyfus is most responsible for this portrayal of AI research in the '50s and '60s. He made a career of insisting that advances in AI would never come to pass, famously predicting that computers couldn't become good at chess because it required "insight", and routinely listing off what he believed were uniquely human qualities that couldn't be embodied in computers, always in the form of underdefined terms such as intuition, insight, and other such terms. Many of the things AI does now are exactly the type of things that doomsayers explicitly predicted would never happen, because they extrapolated from limited progress in the short term to absolute declarations over the infinite timeline of the future. There's a difference between the outer limits of theoretical possibility on the one hand, and instant results over the span of a couple new cycles, and it's unfortunate that these get conflated.
    - RandomLensman 3y ago
      There is no doubt that some leading AI proponents in the 1960s were overconfident.
nly 3y ago
Parrots are pretty intelligent. Seems like a an unfair analogy
IanCal 3y ago
> LLMs won't get intelligent. That's a fact based on their MO. They are sequence completion engines. A system that could perfectly predict what I would do in response to any particular stimuli, as a continuing sequence, would be exactly as intelligent as me. > They can be fine tuned to specific tasks, but at their core, they remain stochastic parrot Othello GPT was an attempt at answering this exact question, it's a simplified setup and appears to learn a world model: https://thegradient.pub/othello/ https://thegradient.pub/othello/
- lewhoo 3y ago
  A system that could perfectly predict what I would do in response to any particular stimuli, as a continuing sequence, would be exactly as intelligent as me. That's certainly interesting but it's not a depiction of a LLM is it ? LLM's are not deterministic, and (perhaps) so are we so two non-deterministic systems can only occasionally align (or so I assume). Intuition says they may get "close enough", whatever that might be, and close enough is good enough in this case but I think you are making a giant assumption to the likes of since we can speed up matter to 1000km/h then IF we sped it up to light speed then ...[something]...
  - stormfather 3y ago
    LLMs are deterministic if the temperature parameter is set to 0. Randomness is artificially injected into their outputs otherwise in order to make them more interesting, but they're just a series of math operations.
    - Philpax 3y ago
      To elucidate on this: the LLM can be viewed as a function that takes the context as an attachment and produces a probability distribution for the next token over all known tokens. Most inference samples from that distribution using a composition of sampling rules or such, but there's nothing stopping you from just always taking the most probable token (temperature = 0) and being fully deterministic. The results are quite bland, but it's perfect for extraction tasks. (note: GPT-4 is not fully deterministic; there's no details on this but the running theory is it is a mixture of experts model and that their expert routing algorithm is not deterministic/is dependent on the resources available)
stared 3y ago
> LLMs won't get intelligent. That's a fact based on their MO. They are sequence completion engines. They can be fine tuned to specific tasks, but at their core, they remain stochastic parrots. Yet, we are different, right?
martindbp 3y ago
If something stays in motion and has been so for some time, it's more important to explain why it would not continue rather than the default assumption that it will stop instantaneously. Show me a curve of diminishing returns and I'll believe you. If an object is in motion you'd need to show me that there is deceleration, or there is a wall just up ahead. But the fact is that the loss goes down predictably with increased compute budget, data and model size (see Chinchilla Scaling Law). We've also seen that decreased loss suddenly results in new capabilities in discontinuous jumps. There is all reason to believe there is still some juice left in this scaling, exactly how far it can be taken is difficult to tell.
czbond 3y ago
> stochastic parrots I mean, let's be honest, so are enough of the bell curve humanity - so LLM's don't need to be amazing. They need to chain together communication that makes them seem sentient (as now) & then be exposed to smaller data sets with specialized, higher level knowledge. This is how humans are... and the reason some are smarter than others.
tempodox 3y ago
> I want to know only one thing, what gives the confidence to say otherwise? Remember that UFO poster? “I want to believe”.
naasking 3y ago
> LLMs won't get intelligent. That's a fact based on their MO. They are sequence completion engines. Where's the proof that sequence completion engines can't be intelligent?

rhn_mk1 3y ago

It's countered by not making the assertion and not being able to make conclusions. You only need a lack of confidence for that. It's orders of magnitude easier to not have knowledge compared to having it.

beefield 3y ago

> LLMs won't get intelligent. That's a fact based on their MO. I kind of agree. However, I see a real possibility that in the near future LLM behaviour would be practically indistinguishable from intelligent/sentient behavior. And at that point we (or at least I) are facing some really interesting/difficult questions, namely how do you know an intelligent looking thing actually is intelligent (or sentient). How do you prove me you/LLM are/aren't a philosophical zombie? How we are supposed to treat very much intelligent/sentient looking things when we are not sure if they are sentient/intelligent or not? Let's face it, lots of people are dumb as rock (too often very much me included). Why we should be able to treat something badly just because we think we know they can't be intelligent, even if they walk , look and quack like intelligent duck? I personally have started to think that the behavior of humans should be judged by the behaviour, not the target. If you want to behave like an asshole towards a teddy bear, then you most likely are an asshole.

savolai 3y ago
It seems to me that the real question here is what is true human intelligence. Ai has made it plain to see, by being able to replicate it so convincingly, that much of what we have considered intelligence has been pattern matching or acting as complex parrots. There is much more to the abilities of human body-mind-emotional-experiential being, but it is only slowly becoming mainstream. (Edit: Of course there are also many analytical skills that AI cannot match at this point. My point is that we shouldn’t overlook any area of human capacity.) One salient question in this is: will we reach a level of intelligence where we become beings capable of actual collaboration that doesn’t waste so much effort in conflicts, or one that is capable of living in harmony within its environment? What capabilities of awareness, trauma work, emotional maturity and self reflection does this require? What resources hidden inside humanity that we have forgotten do we need to wield? Does AI have something to contribute to this process happening?
- worrycue 3y ago
  > It seems to me that the real question here is what is true human intelligence. IMHO the main weakness with LLMs is they can’t really reason. They can statistically guess their way to an answer - and they do so surprisingly well I will have to admit - but they can’t really “check” themselves to ensure what they are outputting makes any sense like humans do (most of the time) - hence the hallucinations.
  - NumberWangMan 3y ago
    Apparently GPT-4 is getting pretty good at knowing when it's wrong: https://thezvi.substack.com/p/ai-26-fine-tuning-time#%C2%A7gpt-real-this-time https://thezvi.substack.com/p/ai-26-fine-tuning-time#%C2%A7g... (They asked GPT-3.5 and GPT-4 "are you sure" to see if it would change its answer, both when the original answer was right, and when it was wrong)
    - worrycue 3y ago
      Does it do that because it can check it’s own reasoning? Or is it just doing so because OpenAI programmed it to not show alternative answers if the probability of the current answer being right is significantly higher than the alternatives?
      - NumberWangMan 3y ago
        I don't know. I don't think anyone is directly programming GPT-4 to behave in any way, they're just training it to give the responses they want, and it learns. Something inside it seems to be figuring out some way of representing confidence in its own answers, and reacting in the appropriate way, or perhaps it is checking its own reasoning. I don't think anyone really knows at this point.
  - naasking 3y ago
    As the other poster said, they can check themselves but this requires an iterative process where the output is fed back in as input. Think of LLMs as the output of a human's stream of consciousness: it is intelligent, but has a high chance of being riddled with errors. That's why we iterate on our first thoughts to refine them.
velvetz 3y ago
Off-topic and might sound strange but I find it intriguing that you wrote behavior and behaviour in these two different forms in the same sentence:)
- coldtea 3y ago
  Quite easy to happen if like many of us you were taugh British English, and then you remember that you're on a US forum, and everybody around you uses the US spelling for things (and you get to read the US variants all the time in other comments).
- beefield 3y ago
  I don't have any excuses but being non native in english having exposure to both British and US english seems to confuse my brain.
js8 3y ago
Any sufficiently advanced intelligence is indistinguishable from an LLM? Cute, but practically speaking, I would prefer the former.

albertzeyer 3y ago

Look at the scaling laws. We found that extrapolating the performance given a few data points with smaller models is actually very accurate. That's how they determined hyper parameters, by tuning them on multiple smaller scale models and then extrapolating. So far, all those predictions were quite good. Together with a bigger model, we also need more data to get better performance. If we add video and audio to the text data, we have still a lot more data we can use, so this is also not really a problem. It would be very unexpected that those scaling laws are suddenly not true anymore for the next order of magnitude in model and data size.

dontupvoteme 3y ago
Scaling laws apply to a single model. The best single model right now is supposedly a 8x mixture of experts, so not even really a single model in the purist sense. I still expect the final solution will be more along the lines of picking the best model(s) from a sea of possible models, switching them in and out as needed, and then automatically reiterating as needed.

golol 3y ago

>LLMs won't get intelligent. That's a fact based on their MO. They are sequence completion engines. They can be fine tuned to specific tasks, but at their core, they remain stochastic parrots. This is absolutely wrong. There is nothing about their MO that stops them from being intelligent. Suppose I build a human LLM as follows: A random human expert is picked and he is shown the current context window. He is given 1 week to deliberate and then may choose the next word/token/character. Then you hook this human LLM into an auto-GPT style loop. There is no reason it couldn't operate with high intelligence on text data. Not also that LLMs are not really about language at all anymore, the architectures can be used on any sequence data. Right now we are compute limited. If compute was 100x cheaper we could have GPT-6, bring 100x bigger, we could have really large and complex agents using GPT-4 power models, or we could train on tupled text-video data of subtitles videos. Given the world model LLMs manage to learn out of text data, I am 100% certain that a sufficiently large transformer can learn a decent world model from text-video data. Then our agents could also have a good physical understanding.

ambrozk 3y ago
Humans will never be intelligent. They're optimized for producing offspring, not reasoning. Humans may appear to be intelligent from a distance, but talk to one for any length of time and you'll find they make basic errors of reasoning that no truly thinking being would fall for. /s
- ksaho 3y ago
  Take out the /s tag and you are right on the money. Humans can not be trusted with anything because they are trivially fallible. Humans are terribly stupid, destroy their own societies and refuse to see reason. They also hallucinate when their destructive tendencies start catching up to them.
  - ambrozk 3y ago
    If the most intelligent machines ever observed in the universe do not count as "intelligent," then we have a semantic, and not a substantive difference of opinion.

squeaky-clean 3y ago

> LLMs won't get intelligent. I think this sentence doesn't mean much unless we have a strict definition of what intelligence means. Just today ChatGPT helped me solve a DNS issue that I would not have been able to solve on my own in one day, let alone an hour. I'd consider it already more intelligent than myself when it comes to DNS.

Aperocky 3y ago
It's seen more DNS content than you and anybody else have seen in their entire lives, and are able to regurgitate what it read because it has far faster memory access than you did. A dictionary contain knowledge but no intelligence.
- TeMPOraL 3y ago
  And LLMs are the opposite of dictionary, actually. They suck at storing facts. They excel at extracting patterns from noise and learning them. It's not obvious to me that this isn't intelligence; on the contrary, I feel it's very much a core component of it.
  - BartjeD 3y ago
    Isn't that a dictionary of patterns?
  - worrycue 3y ago
    > They excel at extracting patterns from noise and learning them. You can argue those are facts too.
    - lucubratory 3y ago
      Yes, I have noticed that a lot of extreme AI cynics have been arguing that any and every example of reasoning or thinking that an LLM displays is just some variant of memorisation.
      - 3y ago

edf13 3y ago

> They are sequence completion engines But at the basic level - isn't our own brain just a sequence completion engine too?

HPsquared 3y ago
The brain does seem to do a lot of pattern-matching and prediction.

nopinsight 3y ago

Geoffrey Hinton, Andrew Ng, and quite a few other top AI researchers believe that current LLMs (and incoming waves of multimodal LFMs) learn world models; they are not simply 'stochastic parrots'. If one feeds GPT-4 a novel problem that does not require multi-step reasoning or very high precision to solve, it can often solve it.

auggierose 3y ago
Anyone who has worked a bit with a top LLM thinks that they learn world models. Otherwise, what they are doing would be impossible. I've used them for things that are definitely not on the web, because they are brand new research. They are definitely able to apply what they've learnt in novel ways.
- RandomLensman 3y ago
  If they learn world models, those world models are incredible poor, i.e., there is no consistency of thought in those world models. In my experience, things outside coding quickly devolve into something more like "technobabble" (and in coding there is always a lot of made-up stuff that doesn't exists in terms of functions etc.).
  - coldtea 3y ago
    >If they learn world models, those world models are incredible poor, i.e., there is no consistency of thought in those world models Incredibly poor compared to ours, but thousands of times better than what "AI" we had before.
    - RandomLensman 3y ago
      Not sure that matters much as they are only for low risk stuff without skilled supervision, so back to advertising, marketing, cheap customer support, etc.
  - auggierose 3y ago
    I see them more as creative artists who have very good intuition, but are poor logicians. Their world model is not a strict database of consistent facts, it is more like a set of various beliefs, and of course those can be highly contradictory.

roenxi 3y ago

I'm put in mind of the OpenAI DoTA bot that was winning 99% of its games and some people refused to admit that it knew how to play DoTA based on some esoteric interpretation of the word "play". We're going to see exponential increases in processing power of the best GPU clusters and human brains are a stationary target. And there is precious little evidence that the average human is much more than an LLM. LLMs are already more likely to understand a topic to a high standard than a given human. They're going to progress and if they aren't intelligent then intelligence is overrated and I'd rather have whatever they have.

alpaca128 3y ago

> LLMs won't get intelligent Even assuming that is true: LLMs aren't all that exists in AI research and just like LLMs are amazing in terms of language it's possible similar breakthroughs could be made in more abstracted areas that could use LLMs for IO. If you think ChatGPT is nice, wait for ChatGPT as frontend for another AI that doesn't have to spend a single CPU cycle on language.

oceanplexian 3y ago
The next AI wave hasn't even started. Imagine an LLM the size of GPT-4 but it's trained on nothing but gene sequence completion. All the models being used in academia are basically toys, none of those guys are running hardware at a scale that can even remotely touch Azure, Meta, etc, and right now there is a massive global shortage of GPU compute that's eventually going to clear up. We know models get A LOT better when they are scaled up and are fed more data, so why wouldn't the same be true for other problems besides text completion?
- td118 3y ago
  [dead]
benterix 3y ago
> LLMs aren't all that exists in AI research Frankly, I'm a bit worried about all the rest now that LLMs proved to be so successful. We might exploit them and arrive to a dead end. In the meantime, other potentially crucial developments in AI might get less attention and funding.

glenstein 3y ago

>Which is countered by...the assertion that it won't? No it's countered by principled restraint in not making an affirmative claim one way or the other. I've heard this referred to as the overconfident pessimism problem. Which is that normal, well founded scientific discipline and evidence-based restraint go out of the window when people declare, without evidence that they know certain advances won't happen. Because people get mentally trapped into this framing of either have to declare that it will happen or that it won't, seeming to forget that you can just adopt the position of modesty and say the dust hasn't yet settled.

tlarkworthy 3y ago
AI has been around the corner since the 1950s, this is the historical evidence for the pessimistic stance against over optimistic predictions. LLMs are a huge stride forward, but AI does not progress like Moore's law. LLM have revealed a new wall. Combining multi agents is not working out as hoped.
- glenstein 3y ago
  Perhaps without intending to, you've cited a pretty appropriate example of overconfident pessimism. Philosopher Hubert Dreyfus is most responsible for this portrayal of AI research in the '50s and '60s. He made a career of insisting that advances in AI would never come to pass, famously predicting that computers couldn't become good at chess because it required "insight", and routinely listing off what he believed were uniquely human qualities that couldn't be embodied in computers, always in the form of underdefined terms such as intuition, insight, and other such terms. Many of the things AI does now are exactly the type of things that doomsayers explicitly predicted would never happen, because they extrapolated from limited progress in the short term to absolute declarations over the infinite timeline of the future. There's a difference between the outer limits of theoretical possibility on the one hand, and instant results over the span of a couple new cycles, and it's unfortunate that these get conflated.
  - RandomLensman 3y ago
    There is no doubt that some leading AI proponents in the 1960s were overconfident.

nly 3y ago

Parrots are pretty intelligent. Seems like a an unfair analogy

IanCal 3y ago

> LLMs won't get intelligent. That's a fact based on their MO. They are sequence completion engines. A system that could perfectly predict what I would do in response to any particular stimuli, as a continuing sequence, would be exactly as intelligent as me. > They can be fine tuned to specific tasks, but at their core, they remain stochastic parrot Othello GPT was an attempt at answering this exact question, it's a simplified setup and appears to learn a world model: https://thegradient.pub/othello/ https://thegradient.pub/othello/

lewhoo 3y ago
A system that could perfectly predict what I would do in response to any particular stimuli, as a continuing sequence, would be exactly as intelligent as me. That's certainly interesting but it's not a depiction of a LLM is it ? LLM's are not deterministic, and (perhaps) so are we so two non-deterministic systems can only occasionally align (or so I assume). Intuition says they may get "close enough", whatever that might be, and close enough is good enough in this case but I think you are making a giant assumption to the likes of since we can speed up matter to 1000km/h then IF we sped it up to light speed then ...[something]...
- stormfather 3y ago
  LLMs are deterministic if the temperature parameter is set to 0. Randomness is artificially injected into their outputs otherwise in order to make them more interesting, but they're just a series of math operations.
  - Philpax 3y ago
    To elucidate on this: the LLM can be viewed as a function that takes the context as an attachment and produces a probability distribution for the next token over all known tokens. Most inference samples from that distribution using a composition of sampling rules or such, but there's nothing stopping you from just always taking the most probable token (temperature = 0) and being fully deterministic. The results are quite bland, but it's perfect for extraction tasks. (note: GPT-4 is not fully deterministic; there's no details on this but the running theory is it is a mixture of experts model and that their expert routing algorithm is not deterministic/is dependent on the resources available)

stared 3y ago

martindbp 3y ago

If something stays in motion and has been so for some time, it's more important to explain why it would not continue rather than the default assumption that it will stop instantaneously. Show me a curve of diminishing returns and I'll believe you. If an object is in motion you'd need to show me that there is deceleration, or there is a wall just up ahead. But the fact is that the loss goes down predictably with increased compute budget, data and model size (see Chinchilla Scaling Law). We've also seen that decreased loss suddenly results in new capabilities in discontinuous jumps. There is all reason to believe there is still some juice left in this scaling, exactly how far it can be taken is difficult to tell.

czbond 3y ago

> stochastic parrots I mean, let's be honest, so are enough of the bell curve humanity - so LLM's don't need to be amazing. They need to chain together communication that makes them seem sentient (as now) & then be exposed to smaller data sets with specialized, higher level knowledge. This is how humans are... and the reason some are smarter than others.

tempodox 3y ago

> I want to know only one thing, what gives the confidence to say otherwise? Remember that UFO poster? “I want to believe”.

naasking 3y ago

> LLMs won't get intelligent. That's a fact based on their MO. They are sequence completion engines. Where's the proof that sequence completion engines can't be intelligent?