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

Google DeepMind CEO says some form of AGI possible in a few years

wsj.com

Post reply on HN
Print view View original

91–100 of 388 posts

Re: Google DeepMind CEO says some form of AGI possible in a few years

#91

Post by garbagecoder »

valine wrote: post #39
TheDudeMan wrote: post #26

Earlier quoted context omitted.

I think we'll have full-blown AGI before fusion and before world-changing levels of quantum computing. Which is to say, we won't be around to see fusion and world-changing quantum :( But there's still time for robo-taxis!

Or, more optimistically, AGI will accelerate scientific progress and we will have fusion, quantum computers, and self driving cars shortly after. If AIs had the tendency to recursively devour worlds you would expect at least one alien AI to have conquered our galaxy by now. But the sky is quiet, so either we are the very first species to get this far, or there’s a natural ceiling to the pace of technological progress…

>you would expect

There is no evidence that an actually existing thing like a Von Neumann probe can survive interstellar travel. I think that the answer to all of these paradoxes is that special relativity just makes it basically impossible. That’s anathema to us nerds, I know.

That’s a very thin argument to hang a Pandora’s Box opening on. It only needs to devour one planet, though we appear to be doing that ourselves too.

Re: Google DeepMind CEO says some form of AGI possible in a few years

#92

Post by NumberWangMan »

matteoraso wrote: post #40

I don't think there's been any progress on AGI, actually. Even though there's a lot of powerful AI programs out there, they're really nothing more than complex tools created to solve problems. A few other commenters mentioned that there's no good definition for what AGI would actually mean, but a good starting point that I think we can all agree on is that an AGI would need to have some level of autonomy. That being…

I don't think that doing the math by hand to simulate a machine intelligence makes it not intelligent, any more than doing a simulation of all the electrochemical signals in a human brain by hand would make a human not intelligent. Aside from the infeasible amount of time it would take, it's the same thing.

As for autonomy, LLMs don't have autonomy by themselves. But they can be pretty easily combined with other systems, connected to the outside world, in a way that seems pretty darned autonomous to me. (https://github.com/Significant-Gravitas/Auto-GPT). And that's basically a duct-tape-and-string version. Given how new these LLMs are, it's likely we're barely scratching the surface.

Re: Google DeepMind CEO says some form of AGI possible in a few years

#93

Post by NumberWangMan »

NumberWangMan wrote: post #92
matteoraso wrote: post #40

I don't think there's been any progress on AGI, actually. Even though there's a lot of powerful AI programs out there, they're really nothing more than complex tools created to solve problems. A few other commenters mentioned that there's no good definition for what AGI would actually mean, but a good starting point that I think we can all agree on is that an AGI would need to have some level of autonomy. That being…

I don't think that doing the math by hand to simulate a machine intelligence makes it not intelligent, any more than doing a simulation of all the electrochemical signals in a human brain by hand would make a human not intelligent. Aside from the infeasible amount of time it would take, it's the same thing. As for autonomy, LLMs don't have autonomy by themselves. But they can be pretty easily combined with other syst…

Also:

> Even though there's a lot of powerful AI programs out there, they're really nothing more than complex tools created to solve problems.

That's what they're designed to be, yes, but sometimes you end up with something more than what you intended.

Re: Google DeepMind CEO says some form of AGI possible in a few years

#94

Post by majikaja »

xbmcuser wrote: post #45
majikaja wrote: post #15

Earlier quoted context omitted.

>we just see a cooling off of other endeavors as LLM becomes a known in terms of its strengths and weaknesses and it’s abilities reach a natural limit which is still general useful. This is my prediction If the training set is just the internet, it will be not much different than someone who spends all their lives in their room in front of their computer.

I don't get all the negativity unlike most humans LLM can be made to discard bad or incorrect data. Quickly transition to new ways of doing something. A human spending all their life in a room in front of the computer would not can not go through all the millions of scientific papers and journals. They would not look at unlimited amount of code or go through all the past and present law of every country . No human is…

I'm not saying it's not useful. I'm saying that once it learns the content of the data sources that are fed to it, that's it. It will spit out some overlooked connections then stop. Maybe the cure for cancer can be found just by reading in between the lines of millions of old scientific papers but I doubt it.

Re: Google DeepMind CEO says some form of AGI possible in a few years

#95

Post by VirusNewbie »

valine wrote: post #39
TheDudeMan wrote: post #26

Earlier quoted context omitted.

I think we'll have full-blown AGI before fusion and before world-changing levels of quantum computing. Which is to say, we won't be around to see fusion and world-changing quantum :( But there's still time for robo-taxis!

Or, more optimistically, AGI will accelerate scientific progress and we will have fusion, quantum computers, and self driving cars shortly after. If AIs had the tendency to recursively devour worlds you would expect at least one alien AI to have conquered our galaxy by now. But the sky is quiet, so either we are the very first species to get this far, or there’s a natural ceiling to the pace of technological progress…

Why do we need the G to do any of these things? I assume non general intelligence should be further along, so shouldn’t this recursive self improvement happen before the general part?

Re: Google DeepMind CEO says some form of AGI possible in a few years

#96

Post by NumberWangMan »

nextos wrote: post #58
soulofmischief wrote: post #46

Earlier quoted context omitted.

Ultimately you are just a system who maintains homeostasis by modeling your environment and reprojecting it to find patterns which can be used to make actions with predictable outcomes. Is this intelligence? Is there only one correct way to achieve this goal? What do you believe intelligence to be?

There's a good summary in the Introduction to Kevin Murphy's second tome [1]. He quotes Josh Tenenbaum, who said: "Intelligence is not just about pattern recognition and function approximation. It’s about modeling the world" . I agree, and I think Goodman & Tenennaum [2] is a great place to see other things that may pop up in the road to AGI. LLMs are great, but they do too much at once. I think moving towards AGI re…

Do we know for sure that LLMs can't possibly be doing symbolic reasoning of some sort, internally? I thought we couldn't really interpret them yet.

Re: Google DeepMind CEO says some form of AGI possible in a few years

#97

Post by AndrewKemendo »

paxys wrote: post #30
ckemere wrote: post #23

Several comments have touched on this, but wouldn't you define an AGI as something that could learn to drive a car? I think every (reasonable ( person understands self driving is >10 years away. So shouldn't AGI be?

Not necessarily. It could be that AGI exists and can learn how to drive a car, but we don't have good enough cameras/LIDARs/other sensors to make it work. Or the computation/battery power needed to do so isn't possible to fit in a car. AGI is about inventing the brain, while the car is an entire body.

In the General case* there is no functional brain without a functional body

*One can nitpick yourself into creating a series of examples of fictional disabled people but let’s not bikeshed please

Re: Google DeepMind CEO says some form of AGI possible in a few years

#98

Post by dragonwriter »

AndrewKemendo wrote: post #97
paxys wrote: post #30

Earlier quoted context omitted.

Not necessarily. It could be that AGI exists and can learn how to drive a car, but we don't have good enough cameras/LIDARs/other sensors to make it work. Or the computation/battery power needed to do so isn't possible to fit in a car. AGI is about inventing the brain, while the car is an entire body.

In the General case* there is no functional brain without a functional body *One can nitpick yourself into creating a series of examples of fictional disabled people but let’s not bikeshed please

Disabled people who have functional brains still have substantial body function (otherwise, their brain – which is part of the body and dependent on a number of other parts – would stop functioning.)

Re: Google DeepMind CEO says some form of AGI possible in a few years

#99

Post by paxys »

AndrewKemendo wrote: post #97
paxys wrote: post #30

Earlier quoted context omitted.

Not necessarily. It could be that AGI exists and can learn how to drive a car, but we don't have good enough cameras/LIDARs/other sensors to make it work. Or the computation/battery power needed to do so isn't possible to fit in a car. AGI is about inventing the brain, while the car is an entire body.

In the General case* there is no functional brain without a functional body *One can nitpick yourself into creating a series of examples of fictional disabled people but let’s not bikeshed please

Can a blind person drive a car? Is a blind person not intelligent?

Re: Google DeepMind CEO says some form of AGI possible in a few years

#100

Post by blast »

PheonixPharts wrote: post #41

As someone who has worked in the field of AI/ML for quite awhile now, the problem with current AGI predictions is ML hasn't done anything new since the 80s (or arguably earlier). At the end of the day all ML is using gradient descent to do some sort of non-linear projection of the data on to a latent space, then doing some relatively simple math in this latent space to perform some task. Personally I think the limits…

"We dont have better algorithms, we just have more data" - Peter Norvig
Post reply on HN