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

Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

venturebeat.com

Post reply on HN
Print view View original

151–160 of 181 posts

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#151

Post by YeGoblynQueenne »

mindgam3 wrote: post #54

Demis Hassabis (true) statements here would be much more credible if DeepMind wasn't currently making a mint by promoting AlphaZero to the masses as a "general purpose artificial intelligence system". Don't believe me? Check out this series of marketing videos on YouTube by GM Matthew Sadler. 1. “Hi, I’m GM Matthew Sadler, and in this series of videos we’re taking a look at new games between AlphaZero, DeepMind’s gen…

Edit: You don't have to go far to see that DeepMind are pushing AlphaZero as "general purpose". This is the title of their Science paper on it:

A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play

http://science.sciencemag.org/content/362/6419/1140

"General" as in what? As opposed to reinforcement learning, in er, general? As opposed to other ANN architectures?

>> I am in no way dismissing AlphaZero's truy remarkable abilities in both chess and other games like go and shogi.

More to the point- it's only chess, go and shogi; not games "like" those.

The AlphaZero architecture has the structure of a chessboard and the range of moves of pieces in chess, go and shogi hard-coded and you can't just take a trained AlphaZero model and apply it to a game that doesn't have either the board or the moves of those three games.

To be blunt, AlphaZero has mastered chess, go and shogi, but it can't play noughts-and-crosses.

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#152

Post by Retric »

joycian wrote: post #140
visarga wrote: post #124

Earlier quoted context omitted.

A rat can do something else that a neural net can't - it is a self replicator. Our neural nets don't have self replication or a huge, complex environment and timescale to evolve in. Self replication creates an internal goal for agents: survival. This drives learning. Instead, we just train agents with human-made rewards signals. Even a simple environment, like the Go board, when used for training many generations of…

Survival is instrumental to any goal. Not only self replication would create that drive.

Bombs don’t need to survive to be useful.

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#153

Post by nradov »

mindcrime wrote: post #136
nradov wrote: post #121

Earlier quoted context omitted.

I ask for proof that AGI is possible. Show me a computer as smart as a lab mouse and then I'll take your concerns seriously. The analogy to anthropomorphic global climate change is a non sequitur. Climatologists have created falsifiable theories which make testable predictions. And you really have no clue about my personal religious beliefs. Calm down and take a seat.

I ask for proof that AGI is possible. Show me a computer as smart as a lab mouse and then I'll take your concerns seriously. I would argue that, unless you can show why AGI is not - in principle - possible, that the null hypothesis would be that it is possible. Unless we veer off into some weird mysticism, it seems that the human brain turns energy and matter into intelligence somehow, operating according to the phys…

You can't prove a negative. At this point we don't even know what the principles are.

If you're unwilling to provide me with a prototype equivalent to a rodent mind then I'll settle for a fully developed theory of human cognition. Let me know when you've got one. At least that would give researchers some guidelines to know whether they're making forward progress toward AGI.

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#154

Post by nradov »

nuguy wrote: post #143
nradov wrote: post #121

Earlier quoted context omitted.

I ask for proof that AGI is possible. Show me a computer as smart as a lab mouse and then I'll take your concerns seriously. The analogy to anthropomorphic global climate change is a non sequitur. Climatologists have created falsifiable theories which make testable predictions. And you really have no clue about my personal religious beliefs. Calm down and take a seat.

I never said anything about your religious beliefs, only that your need for hard proof one way or the other is similar to the need displayed by people who want to believe in god and who dont want to believe in global warming. there are many such examples but the bottom line is that demanding me to show you a sentient computer right now is not reasonable. We have the ability to reason about things without developing f…

No that's not how it works. Currently a belief in the inevitable arrival of AGI is a secular religious faith with no basis in hard science. We have plenty of real problems to worry about (like anthropomorphic global climate change) before we waste time making public policy to prevent something that may never happen anyway. Your concerns are silly, akin to worrying about an alien invasion when we don't even know if extraterrestrial life exists.

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#155

Post by spot »

YeGoblynQueenne wrote: post #149
edanm wrote: post #144

Earlier quoted context omitted.

> AI, brain science and cognitive science are extremely difficult fields with small advances, yet people assume that it's possible to 'brute force' AGI by just adding more computing power and doing more of the same. Do people assume that? I mean, I'm sure some people do, but I don't think I've encountered many people, at least not in the AI safety movement, that actually think it's a matter of more hardware power. So…

That's pretty much the Singularity conjecture in a nutshell: that exponential advances in computing power will drive an exponential increase in machine intelligence. It gets more nuanced than that but there are actually very specialised people who argue very forcefully that AGI is a hair's breadth away and we must act now to protect ourselves from it. Edit: so not "most" people but definitely some very high-profile p…

nope https://en.wikipedia.org/wiki/Technological_singularity#Algo...

"Carl Shulman and Anders Sandberg suggest that algorithm improvements may be the limiting factor for a singularity because whereas hardware efficiency tends to improve at a steady pace, software innovations are more unpredictable and may be bottlenecked by serial, cumulative research."

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#156

Post by mikhailfranco »

Great interview with Hassabis from the BBC. It's meanderingly biographical, with insights about his path through internships, curiosity, startups, commitment, burnout, trusted team mates and eventual successes ...

https://www.bbc.co.uk/sounds/play/p06qvj98

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#157

Post by 995533 »

mortivore wrote: post #28
izzydata wrote: post #24

I'm not even convinced that a real AI is possible with conventional computer hardware or anything remotely similar to it. Not even considering software I get the impression there is a fundamental limitation of hardware.

I'm not convinced we've even defined the problem space well enough to solve it. Like what is the concrete measure(something to target) for intelligence? If we develop general intelligence is it going to be human, dog, or fish?

Shane Legg (co-founder DeepMind) and Marcus Hutter (Schmidhuber pedigree) defined machine intelligence in this canonical paper from 2007: https://arxiv.org/abs/0712.3329

> Universal Intelligence: A Definition of Machine Intelligence

> A fundamental problem in artificial intelligence is that nobody really knows what intelligence is. The problem is especially acute when we need to consider artificial systems which are significantly different to humans. In this paper we approach this problem in the following way: We take a number of well known informal definitions of human intelligence that have been given by experts, and extract their essential features. These are then mathematically formalised to produce a general measure of intelligence for arbitrary machines. We believe that this equation formally captures the concept of machine intelligence in the broadest reasonable sense. We then show how this formal definition is related to the theory of universal optimal learning agents. Finally, we survey the many other tests and definitions of intelligence that have been proposed for machines.

General intelligence usually meant in relation to humans, but you are correct in noting that it is a spectrum, not a binary.

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#158

Post by 995533 »

nradov wrote: post #73
nuguy wrote: post #68

I take huge offense to this article. They claim that when it comes to AGI, Hinton and Hassabis “know what they are talking about.” Nothing could be further from the truth. These are people who have narrow expertise in one framework of AI. AGI does not yet exist so they are not experts in it, in how long it will be, or how it will work. A layman is just as qualified to speculate about AGI as these people so I find it…

Stating something more times doesn't make it true. Everything you've written is pure speculation, and alarmist at that. There's no proof that AGI is even possible, and if it is possible there's no proof that it will end humanity.

Clearly, AGI-level intelligence is possible, because human brains exist.

So unless you pose that a function has to rely on its materialization (there is something untouchably magic about biological neural networks, and intelligence is not multiple realizable), it should be possible to functionally model intelligence. Nature shows the way.

AGI will likely obsolete humanity. Either depricate it, or consume it (make us part of the Borg collective). Heck, even a relatively dumb autonomous atom bomb or computer virus may be enough to wipe humanity from the face of the earth.

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#159

Post by 995533 »

nuguy wrote: post #68

I take huge offense to this article. They claim that when it comes to AGI, Hinton and Hassabis “know what they are talking about.” Nothing could be further from the truth. These are people who have narrow expertise in one framework of AI. AGI does not yet exist so they are not experts in it, in how long it will be, or how it will work. A layman is just as qualified to speculate about AGI as these people so I find it…

You are correct, in that AI experts may not be the best predictors for AGI. For one, they spend their lives working towards the goal of AGI, so it would require a huge amount of cognitive dissonance for them to say that AGI is impossible or very very far on the horizon.

Philosophers and futurists are better suited to hypothesize an AGI timeline.

But you take it too far by saying it is anyone's game.

Game theory, security, and economic competition makes it impossible to globally ban AI. The incentives to automate the economy (compare AI revolution with industrial revolution) and to weaponize AI (Manhattan Project for intelligence) are just too big. We are already seeing that the US focus on fair and ethical AI puts them at a disadvantage against China and Russia. AGI must require pervasive surveillance of the populace, but the Luddites are holding this back.

I suggest you learn to stop worrying about the bomb, and start planning for its arrival.

Post reply on HN