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

191–200 of 1001 posts

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

#191

Post by ben_w »

AnimalMuppet wrote: post #156
omidsa1 wrote: post #146

Earlier quoted context omitted.

I also quite like the way he puts it. However, from a certain point onward, the AI itself will contribute to the development—adding nines—and that’s the key difference between this analogy of nines in other systems (including earlier domain‑specific ML ones) and the path to AGI. That's why we can expect fast acceleration to take off within two years.

Isn't that one of the measures of when it becomes an AGI? So that doesn't help you with however many nines we are away from getting an AGI. Even if you don't like that definition, you still have the question of how many nines we are away from having an AI that can contribute to its own development. I don't think you know the answer to that. And therefore I think your "fast acceleration within two years" is unsupporte…

AI has been helping with the development of AI ever since at least the first optimising compiler or formal logic circuit verification program.

Machine learning has been helping with the development of machine learning ever since hyper-parameter optimisers became a thing.

Transformers have been helping with the development of transformer models… I don't know exactly, but it was before ChatGPT came out.

None of the initials in AGI are booleans.

But I do agree that:

> "fast acceleration within two years" is unsupported, just wishful thinking

Nobody has any strong evidence of how close "it" is, or even a really good shared model of what "it" even is.

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

#192

Post by michaelt »

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…

> Most of these companies value is built on the idea of AGI being achievable in the near future.

Is it? Or is it based on the idea a load of white collar workers will have their jobs automated, and companies will happily spend mid four figures for tech that replaces a worker earning mid five figures?

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

#194

Post by chasd00 »

awesome_dude wrote: post #49

I have massive respect for Andrej, my first encounter with "him" was following his tutorials/notes when he was a grad student/tutor for AI/ML. I was a lot disappointed when he went to work for Tesla, and I think that he had some achievement there, butnot nearly the impact I believe he potentially has. His switch (back?) to OpenAI was, in my mind, much more in keeping with where his spirit really lies. So, with that i…

if openAI didn't put a chat interface in front of an LLM and make it available to the public wouldn't we still be in the same AI winter? Google, Meta, Microsoft, all of the major players were doing lots of LLM work already, it wasn't until the general public found out through the OpenAI's website that it really took off. I can't remember who said it, it was some CEO, that OpenAI had no moat but nether did anyone else. They all had LLMs already of their own. Was the breakthrough the LLM or making it accessible to the general public?

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

#195

Post by ekjhgkejhgk »

sysguest wrote: post #159
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…

yeah that "model of the world" would mean: babies are already born with "the model of the world" but a lot of experiments on babies/young kids tell otherwise

> yeah that "model of the world" would mean: babies are already born with "the model of the world"

No, not necessarily. Babies don't interact with the world only by reading what people wrote wikipedia and stackoverflow, like these models are trained. Babies do things to the world and observe what happens.

I imagine it's similar to the difference between a person sitting on a bicycle and trying to ride it, vs a person watching videos of people riding bicycles.

I think it would actually be a great experiment. If you take a person that never rode a bicycle in their life and feed them videos of people riding bicycles, and literature about bikes, fiction and non-fiction, at some point I'm sure they'll be able to talk about it like they have huge experience in riding bikes, but won't be able to ride one.

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

#196

Post by cactusplant7374 »

exe34 wrote: post #181
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…

To me, it's a matter of a very big checklist - you can keep adding tasks to the list, but if it keeps marching onwards checking things off your list, some day you will get there. whether it's a linear or asymptotic march, only time will tell.

That's like saying that if we image every neuron in the brain we will understand thinking. We can build these huge databases and they tell us nothing about the process of thinking.

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

#197

Post by CaptainOfCoit »

input_sh wrote: post #138
CaptainOfCoit wrote: post #15

Earlier quoted context omitted.

And a program that can write, sound and paint like a human was 20 years away perpetually as well, until it wasn't.

Another way to put it is that it writes, sounds and paints as the Internet's most average user. If you train it on a bunch of paintings whose quality ranges from a toddler's painting to Picasso's, it's not going to make one that's better than Picasso's, it's going to output something more comparable to the most average painting it was trained on. If you then adjust your training data to only include world's best pain…

> Another way to put it is that it writes, sounds and paints as the Internet's most average user.

Yes, I agree, it's not high quality stuff it produces exactly, unless the person using it already is an expert and could produce high quality stuff without it too.

But there is no denying it that those things were regarded as "far-near future maybe" for a long time, until some people put the right pieces together.

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

#198

Post by breuleux »

omidsa1 wrote: post #146
Imnimo wrote: post #68

>What takes the long amount of time and the way to think about it is that it’s a march of nines. Every single nine is a constant amount of work. Every single nine is the same amount of work. When you get a demo and something works 90% of the time, that’s just the first nine. Then you need the second nine, a third nine, a fourth nine, a fifth nine. While I was at Tesla for five years or so, we went through maybe three…

I also quite like the way he puts it. However, from a certain point onward, the AI itself will contribute to the development—adding nines—and that’s the key difference between this analogy of nines in other systems (including earlier domain‑specific ML ones) and the path to AGI. That's why we can expect fast acceleration to take off within two years.

I don't think we can be confident that this is how it works. It may very well be that our level of intelligence has a hard limit to how many nines we can add, and AGI just pushes the limit further, but doesn't make it faster per se.

It may also be that we're looking at this the wrong way altogether. If you compare the natural world with what humans have achieved, for instance, both things are qualitatively different, they have basically nothing to do with each other. Humanity isn't "adding nines" to what Nature was doing, we're just doing our own thing. Likewise, whatever "nines" AGI may be singularly good at adding may be in directions that are orthogonal to everything we've been doing.

Progress doesn't really go forward. It goes sideways.

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

#199

Post by throwaway-0001 »

mkbelieve wrote: post #114

I don't understand how anyone can believe that we're near even a whiff of AGI when we barely understand what dreaming is, or how the human brain interacts with the quantum world. There are so many elements of human creativity that are still utterly hidden behind a wall that it makes me feel insane when an entire industry is convinced we're just magically going to have the answer soon. The people heralding the emergen…

We Dont know how a horse works, but we got cars. Analogy doesn’t work.

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

#200

Post by Root_Denied »

layer8 wrote: post #142
Root_Denied wrote: post #89

Earlier quoted context omitted.

Except AI already had a clear definition well before it started being used as a way to inflate valuations and push marketing narratives. If nothing else it's been a sci-fi topic for more than a century. There's connotations, cultural baggage, and expectations from the general population about what AI is and what it's capable of, most of which isn't possible or applicable to the current crop of "AI" tools. You can't j…

Maybe do some reading here: https://en.wikipedia.org/wiki/History_of_artificial_intellig...

And you should do some reading into the edit history of that page. Wikipedia isn't immune from concerted efforts to astroturf and push marketing narratives.

More to the point, the history of AI up through about 2010 talks about attempts to get it working using different approaches to the problem space, followed by a shift in the definitions of what AI is in the 2005-2015 range (narrow AI vs. AGI). Plenty of talk about the various methods and lines fo research that were being attempted, but very little about publicly pushing to call commercially available deliverables as AI.

Once we got to the point where large amounts of VC money was being pumped into these companies there was an incentive to redefine AI in favor of what was within the capabilities and scope of machine learning and LLMs, regardless of whether that fit into the historical definition of AI.

Post reply on HN