Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The changing goalposts of AGI and timelines

mlumiste.com

Post reply on HN
Print view View original

381–390 of 411 posts

Re: The changing goalposts of AGI and timelines

#381

Post by dllthomas »

marcus_holmes wrote: post #348
dragonwriter wrote: post #322

Earlier quoted context omitted.

The model is not the system. The model is a component of the system. The "cron job" (or other means by which a continuous action loop is implemented) and the necessary prompting for it to gather input (including subsequent user input or other external data) and to pursue a set of objectives which evolves based on input are all also parts of the system.

yeah, I get that. But the actual bit that's doing the thinking is restarting from scratch every time. It loads the context, does the next thing, maybe updates the context, shuts down. One second later the same thing. This is not "highly autonomous" Artifical Intelligence. Just IMHO. Other opinions are also valid.

If you learned that a piece of the brain where meaningful computation takes place was stateless, would that cause you to question whether the human mind was "highly autonomous"?

Re: The changing goalposts of AGI and timelines

#382

Post by nomel »

godelski wrote: post #312
nomel wrote: post #306

Earlier quoted context omitted.

Sorry, I assumed the context was cleared, with the article above. Here's what I meant: > If it can do things as good as or better than humans, in general, then either the AI has a type of general intelligence ...

Sorry, I assumed the comment was clear, with your comment above. Here's what I meant: > By your definition every machine has a type of general intelligence. Not just a bog standard calculator, but also my broom. I really don't know of any human that can out perform a standard calculator at calculations. I'm sure there are humans that can beat them in some cases, but clearly the calculator is a better generalized nume…

I'm using the dictionary definition of "general":

> General: 1. affecting or concerning all or most people, places, or things; widespread.

To me, a general intelligence is one that is not just specific: it's affecting or concerning all or most areas of intelligence.

A calculator is a very specific, very inflexible, type of intelligence, so it's not, by definition, general. And, I'm not talking about the indirect applications of a calculator or a specific intelligence.

If you want to argue that we don't need the concept of AGI, because something like specific experts could be enough to drastically change the economy, then sure! That would be true. But I think that's a slightly different, complimentary, conversation. Even then, say we have all these experts, then a system to intelligently dispatch problems to them...maybe that's a specific implementation of AGI that would work. I think how less dependent on human intelligence the economy becomes, and how more dependent on non-human decision makers it becomes, is a reasonable measure. This seems controversial, which I can't really understand. I'm in hardware engineering, so maybe I have a different perspective, but goals based on outcome are the only ones that actually matter, especially if nobody has done it before.

Re: The changing goalposts of AGI and timelines

#383

Post by nomel »

Jensson wrote: post #373
computably wrote: post #286

Earlier quoted context omitted.

I'm pretty sure they were asking for a pinned date for definitions of "economically valuable" and "most (of total economic value)", specifically because, as previous comments noted, the definition and quantity of "economic value" vary over time. If AI hype is to be believed, and if we assume AGI has a slow takeoff, the economy will look very different in 2030, significantly shifting the goalposts for AGI relative to…

Well if humans can do economically valuable mental work the AI can't then its not AGI, don't you think? An AGI could learn that new job too and replace the human, so as long as we still have economically valuable mental work that only humans can do then we haven't reached AGI.

This is a strange binary I don't understand. There are humans that can't do the work of some humans. Intelligence is, clearly, a spectrum. I don't see why a general intelligence would need to have capabilities far beyond a human, when just replacing somewhat lacking humans could upend large portions of the economy. Again, "it's not AGI" arguments will eventually require that some humans aren't considered intelligent, which is the point in time that we'll all be able to agree "ok, this is AGI".

Re: The changing goalposts of AGI and timelines

#384

Post by irishcoffee »

esafak wrote: post #296
irishcoffee wrote: post #233

Earlier quoted context omitted.

Do you know how an LLM works? Can you describe it?

Do you know how the human brain works? That science is still in its infancy but that's not stopped us.

Here is a run-down of how an LLM works:

https://en.wikipedia.org/wiki/Large_language_model

> Do you know how the human brain works? That science is still in its infancy but that's not stopped us.

Stopped us from doing what exactly?

Re: The changing goalposts of AGI and timelines

#385

Post by nomel »

godelski wrote: post #304
tbrownaw wrote: post #289

Earlier quoted context omitted.

As catlifeonmars noted, what's valuable changes over time. But beyond that, part of the nature of that change over time is that things tend to be valuable because they're scarce . So the definition from upthread becomes roughly "highly autonomous systems that outperform humans at [useful things where the ability to do those things is scarce]", or alternatively "highly autonomous systems that outperform humans at [use…

I'd argue it's so vague it's already nonsensical. Can we not declare Google (search) AGI? It sure does a hell of a lot of stuff better than any human I now. Same with the calculator in my desk drawer. Even by broom does a far better job sweeping than I do. My hands just aren't made for sweeping. But to extend your point, I think we really need to be explicit about the assumptions being made. Everyone loves to say int…

I think you're conflating "knowledge" with "intelligence". And, "agency" seems to be a missing concept, which is the only way for something intelligent to apply its knowledge to achieve something practical, on its own.

Google search can't achieve anything practical, because it has no agency. It has no agency partly because it doesn't have the required intelligence to do anything on its own, other than display results for something else, that does have agency, to use.

The applicable definitions, from the dictionary:

Knowledge: facts, information, and skills acquired through experience or education; the theoretical or practical understanding of a subject.

Intelligence: the ability to acquire and apply knowledge and skills.

Agency: the ability to make decisions and act independently.

Re: The changing goalposts of AGI and timelines

#386

Post by nomel »

irishcoffee wrote: post #384
esafak wrote: post #296

Earlier quoted context omitted.

Do you know how the human brain works? That science is still in its infancy but that's not stopped us.

Here is a run-down of how an LLM works: https://en.wikipedia.org/wiki/Large_language_model > Do you know how the human brain works? That science is still in its infancy but that's not stopped us. Stopped us from doing what exactly?

I think a charitable interpretation of their comment is that it hasn't stopped us from practical, fruitful, use.

We didn't need an understanding of how it worked, or even a word for intelligence, let alone a definition of it, to get good practical results.

Re: The changing goalposts of AGI and timelines

#387

Post by dragonwriter »

marcus_holmes wrote: post #348
dragonwriter wrote: post #322

Earlier quoted context omitted.

The model is not the system. The model is a component of the system. The "cron job" (or other means by which a continuous action loop is implemented) and the necessary prompting for it to gather input (including subsequent user input or other external data) and to pursue a set of objectives which evolves based on input are all also parts of the system.

yeah, I get that. But the actual bit that's doing the thinking is restarting from scratch every time. It loads the context, does the next thing, maybe updates the context, shuts down. One second later the same thing. This is not "highly autonomous" Artifical Intelligence. Just IMHO. Other opinions are also valid.

> But the actual bit that's doing the thinking is restarting from scratch every time.

A sibling comment questions the relevance of this by asking what would change if that were true of some low-level component of the human thinking engine, which is a good point, but also: what "actual bit" does this? Both commercial backends and even desktop inference software usually does prefix caching in memory, so arguably that doesn't model the core piece of software running low-level inference, except when either the past context is changed (e.g., when compacting to manage context, or if you are using one software instance and swapping logical histories because you are running multiple agents concurrently — but not in parallel — on one software engine.) And it obviously doesn't match the system at any higher level.

> This is not "highly autonomous" Artifical Intelligence.

Even if that was an accurate model of a component of the system, that a component considered alone is not highly autonomous artificial intelligence is not an argument that the aggregate system is not a highly anonymous artificial intelligence.

Re: The changing goalposts of AGI and timelines

#388

Post by sulam »

ACCount37 wrote: post #364
RugnirViking wrote: post #359

Earlier quoted context omitted.

> I could give you a list of chess moves and force you to recover the complete board state from it, and you wouldn't fare that much better than an off the shelf LLM would idk, I would expect anyone with an understanding of the rules of chess, and an understanding of whatever notation the moves are in, would be able to do it reasonably well? does that really sound so hard to you? people used to play correspondance che…

I would expect that off the shelf GPT-5.4 would be able to do it when prompted carefully, yes. Through reasoning - by playing every move step by step and updating the board one move at a time to arrive at a final board state. On the other hand, recovering the full board state in a single forward pass? That takes some special training. Same goes for meatbag chess. A correspondence chess aficionado might be able to tak…

Try to play a simple over the board style game with 5.4 with whatever notation you chose to use (or just descriptions, literally anything). Prediction: it will start out fine, but the mid game will be very hard to keep it on track, and the endgame will make you give up.

Re: The changing goalposts of AGI and timelines

#389

Post by trollbridge »

tim333 wrote: post #346
neurocline wrote: post #251

Earlier quoted context omitted.

I have to imagine the poster was referring to Dora the Explorer, a popular and charming cartoon from the start of this century.

I think it must be Dory who has short term memory loss https://youtu.be/B6178Ac90S4?t=22

Obviously my long-term memory is not working so well either.

Re: The changing goalposts of AGI and timelines

#390

Post by marcus_holmes »

dllthomas wrote: post #381
marcus_holmes wrote: post #348

Earlier quoted context omitted.

yeah, I get that. But the actual bit that's doing the thinking is restarting from scratch every time. It loads the context, does the next thing, maybe updates the context, shuts down. One second later the same thing. This is not "highly autonomous" Artifical Intelligence. Just IMHO. Other opinions are also valid.

If you learned that a piece of the brain where meaningful computation takes place was stateless, would that cause you to question whether the human mind was "highly autonomous"?

Good question.

I don't know enough about neuroscience to really answer your question in depth.

My opinion, uninformed as it is, is basically around the intuitive reasoning that something cannot be "highly autonomous" if it has to be kicked every second ;) Autonomous is defined as not needing to be controlled externally. And coupling that part with something as simple as cron job doesn't solve that in any meaningful way or make it "autonomous".

A batch file coupled with a cron job that triggers it once a day is not an "autonomous system" to my mind. It's a scheduled system, and there's a significant difference between those things.

Post reply on HN