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

John Carmack's new AGI company, Keen Technologies, has raised a $20M round

twitter.com

Post reply on HN
Print view View original

431–440 of 643 posts

Re: John Carmack's new AGI company, Keen Technologies, has raised a $20M round

#431

Post by Jack000 »

Calavar wrote: post #400
Jack000 wrote: post #327

Earlier quoted context omitted.

Which part do you find objectionable - the lack of progress in previous years or the current/future potential of transformers in RL? I do work in ML but mostly applications instead of research.

One big issue is how exactly we'll continue to scale. Exponential growth is hard to maintain. An example: In the Chinchilla paper [1], the authors suggest that most big transformer models are undertrained, and that we will probably see diminishing returns in scaling up the size of networks if we don't also scale up the size of the datasets. They have a subanalysis where they extrapolate out how big datasets will need…

The scaling laws of large language models are very specific to language models and the way they're trained. The important thing that LLMs demonstrate is that transformers are capable of this kind of scale (where other approaches have not)

In the RL space, a sufficiently complex, stochastic environment is effectively a data generator.

Re: John Carmack's new AGI company, Keen Technologies, has raised a $20M round

#432

Post by labrador »

version_five wrote: post #422
labrador wrote: post #419

Earlier quoted context omitted.

> gullibility gap Search turns up very little except a book by a Christian author. Could you define what you mean by gullibility gap?

It's coined in context.

I don't see it. Did they mean "uncanny valley"?

Re: John Carmack's new AGI company, Keen Technologies, has raised a $20M round

#433

Post by godelski »

darawk wrote: post #423
godelski wrote: post #412

Earlier quoted context omitted.

I'm not sure Gato actually demonstrates intelligence. It definitely does demonstrate the ability to perform a wide variety of tasks, but I don't think that is actually intelligence. Gato has yet to demonstrate that it understands things. I do think it is easy to fall prone to the gullibility gap (and also take too strong of an opposing view, which I may be victim to).

How are you defining "understands things"? Or more precisely, what are the observable consequences of understanding things that you are waiting to see?

"Understands things" means a lot. Mostly I'm referring the causality and generalization of inference though.

Re: John Carmack's new AGI company, Keen Technologies, has raised a $20M round

#434

Post by sigmoid10 »

andreyk wrote: post #413
sigmoid10 wrote: post #403

Earlier quoted context omitted.

As an AI researcher myself I can definitely see why people used to have these sentiments, but I also have to point out how the big papers of the last ~12 months or so have changed the landscape. Especially multi-modal models like Gato. To my surprise, a lot of people (even inside the community) still tend to put human intelligence on some sort of pedestal, but I believe once a system (or host of systems) achieves a s…

Gato is definitely cool, but it's worth remembering it was based on supervised learning with data from a bunch of tasks, so data availability is a strong bottleneck with that approach. Still, I'd agree the past year or two have made me more optimistic about HLAI (if not scary AGI).

Yeah, I'm definitely not saying this is already the ultimate solution. But if you look at their scaling experiments and extrapolate the results, it roughly shows how a model with number of parameters on the same order of magnitude as the human brain could more or less achieve overall human level performance. Of course that doesn't mean the architecture has to extrapolate, but it's definitely a path worth walking unless someone proves it's a dead end. In that case I'd replace "end of decade" with 10 to 20 years in my above statement.

Re: John Carmack's new AGI company, Keen Technologies, has raised a $20M round

#435

Post by kordlessagain »

godelski wrote: post #409
thewarrior wrote: post #390

Earlier quoted context omitted.

I know it maybe too much to ask but as a noob to this field I’d love to hear more about why AGI is not feasible even in 50 years.

Sure. It is rather complex to be honest but I'll do my best to give a high level abstract. I should first off explain that there's no clear definition of intelligence. You might see some of my other comments use the term HLI, which is Human Level Intelligence. It is a term gaining in popularity due to this difficult to define definition and allows for more a porn like definition -- I know it when I see it. Philosophe…

> You can tell a human a pretty fictitious scenario and we're really good at producing those images.

Not all us do that.

A chair is both what I see as well as what I perceive. And it is neither. But, I never build a visual model of it when my eyes aren’t seeing it.

Re: John Carmack's new AGI company, Keen Technologies, has raised a $20M round

#436

Post by godelski »

version_five wrote: post #422
labrador wrote: post #419

Earlier quoted context omitted.

> gullibility gap Search turns up very little except a book by a Christian author. Could you define what you mean by gullibility gap?

It's coined in context.

Really I'd say it is coined by Gary Marcus. At least that's who I first heard discuss it.

Re: John Carmack's new AGI company, Keen Technologies, has raised a $20M round

#437

Post by nradov »

sigmoid10 wrote: post #403
godelski wrote: post #371

Earlier quoted context omitted.

> As someone working in AI research Being in the tail end of my PhD, I want to second this sentiment. I'm not even bullish on AGI (more specifically HLI) in 50 years. Scale will only take you so far and we have to move past frequentism. Hell, causality research still isn't that popular but is quite important for intelligence. I think people (especially tech enthusiasts) are getting caught in the gullibility gap.

As an AI researcher myself I can definitely see why people used to have these sentiments, but I also have to point out how the big papers of the last ~12 months or so have changed the landscape. Especially multi-modal models like Gato. To my surprise, a lot of people (even inside the community) still tend to put human intelligence on some sort of pedestal, but I believe once a system (or host of systems) achieves a s…

If you review the history of AGI research going back to the 1960s, there have been multiple breakthroughs which convinced some researchers that success was close. Yet those all turned out to be mirages. You're probably being overly optimistic, and the Gato approach will eventually run into a wall.

Re: John Carmack's new AGI company, Keen Technologies, has raised a $20M round

#438

Post by godelski »

labrador wrote: post #432
version_five wrote: post #422

Earlier quoted context omitted.

It's coined in context.

I don't see it. Did they mean "uncanny valley"?

Kinda. The gullibility gap more refers to the fact that it is really easy to believe the machines are smarter than they are because they can perform tasks that nothing but humans can do. With the exception of maybe parrots there's no other creatures that can talk so we associate this with intelligence. But I think it is rather naive to believe that the ability to string together reasonable sounding text is actually intelligence. There has yet to be a demonstration of these algorithms actually understanding things.

Re: John Carmack's new AGI company, Keen Technologies, has raised a $20M round

#439

Post by thfuran »

thewarrior wrote: post #390
godelski wrote: post #371

Earlier quoted context omitted.

> As someone working in AI research Being in the tail end of my PhD, I want to second this sentiment. I'm not even bullish on AGI (more specifically HLI) in 50 years. Scale will only take you so far and we have to move past frequentism. Hell, causality research still isn't that popular but is quite important for intelligence. I think people (especially tech enthusiasts) are getting caught in the gullibility gap.

I know it maybe too much to ask but as a noob to this field I’d love to hear more about why AGI is not feasible even in 50 years.

They didn't say it's not possible within 50 years, they said they weren't expecting it to happen in that timeframe, and that's a very different sort of claim.

Re: John Carmack's new AGI company, Keen Technologies, has raised a $20M round

#440

Post by sytelus »

nighthawk454 wrote: post #341
mywittyname wrote: post #219

Earlier quoted context omitted.

Ah, I was thinking that $20MM doesn't seem like a lot of money for someone like Carmack. Surely he could have self-funded a business himself. This explains why he didn't.

yeah, his follow up tweet is pretty clear: > This is explicitly a focusing effort for me. I could write a $20M check myself, but knowing that other people's money is on the line engenders a greater sense of discipline and determination. I had talked about that as a possibility for a while, and I am glad Nat pushed me on it.

It seems like bad logic. If he gets bored/throw in towel, he can just pay back with his money and move on. Also, if you need to create such artificial constraints in order to go to work, a main ingredient called “drive” is missing. I respect Carmack a lot but I don’t trust old centi-millionaires to work with same passion as their younger poorer selves. Finally, $20M is too low. AGI is not a solo endeavor and doesn’t fit in Carmack’s legendary work style. Just to train a large model once will eat up 20% of this money.
Post reply on HN