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

Andrej Karpathy – It will take a decade to work through the issues with agents

dwarkesh.com

Post reply on HN
Print view View original

351–360 of 1001 posts

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#351

Post by lelandbatey »

robotresearcher wrote: post #277
godelski wrote: post #224

Earlier quoted context omitted.

> that to understand knowledge you have to have a model of the world. You have a small but important mistake. It's to recite (or even apply ) knowledge. To understand does actually require a world model. Think of it this way: can you pass a test without understanding the test material? Certainly we all saw people we thought were idiots do well in class while we've also seen people we thought were geniuses fail. The t…

In my view 'understand' is a folk psychology term that does not have a technical meaning. Like 'intelligent', 'beautiful', and 'interesting'. It usefully labels a basket of behaviors we see in others, and that is all it does. In this view, if a machine performs a task as well as a human, it understands it exactly as much as a human. There's no problem of how to do understanding, only how to do tasks. The 'problem' me…

> if a machine performs a task as well as a human, it understands it exactly as much as a human.

I think you're right, except that the ones judging "as well as a human" are in fact humans, and humans have expectations that expand beyond the specs. From the narrow perspective of engineering specifications or profit generated, a robot/AI may very well be exactly as understanding as a human. For the people which interact with those systems outside the money/specs/speeds & feeds, the AI/robot will always feel at least different compared to a person. And as long as it's different, there will always be room to un-falsifiably claim "this robot is worse in my opinion due to X/Y/Z difference."

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#353

Post by sameermanek »

zeroonetwothree wrote: post #220
awongh wrote: post #75

Now that Nvidia is the most valuable company, all this talk of actual AGI will be washed away by the huge amount of dollars driving the hype train. Most of these companies value is built on the idea of AGI being achievable in the near future. AGI being too close or too far away affects the value of these companies- too close and it'll seem too likely that the current leaders will win. Too far away and the level of sp…

It’s possible for AI to provide tremendous economic value without AGI

That is for a governing body to look out for. NOT private companies. Governments have a job to run massive programs for socioeconomic welfare without carrying about profit.

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#354

Post by WJW »

chasd00 wrote: post #336
qlm wrote: post #292

Earlier quoted context omitted.

Language isn't thought. It's a representation of thought.

Something to think about (hah!) is there are people without an internal monologue i.e. no voice inside their head they use when working out a problem. So they're thinking and learning and doing what humans do just fine with no little voice no language inside their head.

It's so weird that people literally seem to have a voice in their head they cannot control. For me personally my "train of thought" is a series of concepts, sometimes going as far as images. I can talk to myself in my head with language if I make a conscious effort to do so, just as I can breathe manually if I want. But if I don't, it's not really there like some people seem to have.

Probably there are at least two groups of people and neither really comprehends how the other thinks haha.

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#356

Post by CamperBob2 »

qlm wrote: post #292
bentt wrote: post #279

Earlier quoted context omitted.

I think this a useful challenge to our normal way of thinking. At the same time, "the world" exists only in our imagination (per our brain). Therefore, if LLMs need a model of a world, and they're trained on the corpus of human knowledge (which passed through our brains), then what's the difference, especially when LLMs are going back into our brains anyway?

Language isn't thought. It's a representation of thought.

If it were that simple, LLMs wouldn't work at all.

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#357

Post by xboxnolifes »

Yizahi wrote: post #316
asah wrote: post #210

Earlier quoted context omitted.

Depends on the definition, I might take that bet because under some definitions were already here. Example: better than average human across many thinking tasks is done.

Calculator can do arithmetic better than a human. Does this mean we have so called AI for half a century now?

A calculator does 1 thinking task.

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#358

Post by guluarte »

jb1991 wrote: post #58

I would bet all of my assets of my life that AGI will not be seen in the lifetime of anyone reading this message right now. That includes anyone reading this message long after the lives of those reading it on its post date have ended. Which of course raises the interesting question of how I can make good on this bet.

my bet is we will just slowly automate things more and more until one day someone will point out when we reached "AGI"

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#359

Post by chasd00 »

benzible wrote: post #213

What's his estimate of how far we are from a definition of AGI?

I wonder if like an inverse definition would work. If a user has root/admin access and all permissions/authority needed to command an AI to list the files in a directory and it simply refuses, would that be a sign of intelligence?

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#360

Post by munksbeer »

godelski wrote: post #224
ekjhgkejhgk wrote: post #134

Earlier quoted context omitted.

The interview which I've watched recently with Rich Sutton left me with the impression that AGI is not just a matter of adding more 9s. The interviewer had an idea that he took for granted: that to understand language you have to have a model of the world. LLMs seem to udnerstand language therefore they've trained a model of the world. Sutton rejected the premise immediately. He might be right in being skeptical here…

> that to understand knowledge you have to have a model of the world. You have a small but important mistake. It's to recite (or even apply ) knowledge. To understand does actually require a world model. Think of it this way: can you pass a test without understanding the test material? Certainly we all saw people we thought were idiots do well in class while we've also seen people we thought were geniuses fail. The t…

> We always had the math to show that scale wasn't enough

Math, to show that scale (presumably of LLMs) wasn't enough for AGI?

This sounds like it would be quite a big deal, what math is that?

Post reply on HN