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

Likelihood of discontinuous progress around the development of AGI

aiimpacts.org

Post reply on HN
Print view View original

91–100 of 111 posts

Re: Likelihood of discontinuous progress around the development of AGI

#91

Post by sgt101 »

abecedarius wrote: post #80
simonh wrote: post #38

Earlier quoted context omitted.

Those vectors are only used to generate the move trees though. That part of the architecture is common to pretty much all MCTS board game AIs ever. The value in AlphaZero is in the neural nets used for the expert policy and the value functions and those don’t have anything about the game rules encoded into them at all. I agree it’s probably quite constrained in the range of possible applications. Everyone was expecti…

> Everyone was expecting Deep Blue to revolutionise AI applications too. That doesn't match my memory at all. The reaction then was dominated by the likes of "this is super-narrow, not real intelligence". (The 80s did have a lot of hyped expectations of related tech, it's true, but that was around 10-15 years earlier.)

I thought Watson would; what's happened to that project is a salient lesson.

Re: Likelihood of discontinuous progress around the development of AGI

#92

Post by sgt101 »

AndrewKemendo wrote: post #45
YeGoblynQueenne wrote: post #29

About AlphaZero particularly, a few things must be kept in mind. First, AlphaZero still makes use of a Monte Carlo Tree Search algorithm to search for good moves. MCTS is a powerful algorithm with a very limited scope: zero-sum, perfect information games. So for instance, it would be very difficult to see how to use MCTS-based AlphaZero in, e.g., training self-driving cars. Second, the AlphaZero architecture is preci…

So, AlphaZero doesn't mean we're closer to _general_ AI. Quite the contrary: it's a very specialised form of AI that will be very difficult to use in any different task than chess, shoggi or go. This is a very true statement and one that I think a lot of people who aren't in ML/DL, but are "worried" about AGI, miss. There is however a common thread with everyone in AI, that they tend to think of AGI as "One algorithm…

Well - SOAR.

Re: Likelihood of discontinuous progress around the development of AGI

#93

Post by tristanm »

AndrewKemendo wrote: post #77
tristanm wrote: post #75

Earlier quoted context omitted.

The "hivemind" argument seems to predict that as society scales up (either through massive population growth or through faster and better interconnectedness, such as through the internet) that as a result we should be seeing much faster gains in technological progress especially in the last few decades or so. However there are quite a few observations that a lot of this progress has sort of slowed down compared to th…

I don't think your interpretation of that prediction is accurate. Rather it would be that there are "bursts" of technological progress followed by slower or no gains while the world "catches up." That seems to more accurately follow the history of technological progress. I think a better interpretation would be that those "bursts" happen at tighter intervals, and if you look at the course of history that seems to be…

I think that bursts of technological progress follows more from the model of individualized intelligence, whereas continuous progress follows from the distributed, networked model of intelligence.

A promoter of the distributed model of intelligence might argue that Einstein was only able to produce the general theory of relativity because of the knowledge already contained within society, such as the mathematics and physics that had already been built up to that time. All the stuff from Euclid to Newton to Gauss to Poincare and Minkowski that Einstein's work relied upon.

Does that imply that Einstein wasn't really smart? If you narrow your focus to just the innovation Einstein made, where did that come from? Did it come from the "hivemind" or was Einstein himself doing something special that allowed him to develop the insight?

More individualized intelligence would predict that we would see smaller intervals between bursts as society increases in size and connectedness (more chances for Einsteins to appear, more likelihood that they can work together). But if intelligence is somehow an emergent process from the network of all humans itself, then as society grows we shouldn't see many bursts at all, just a fairly continuous increase in knowledge as little bits and pieces get absorbed and distributed.

Re: Likelihood of discontinuous progress around the development of AGI

#94

Post by Nomentatus »

"low base rate [of change] for all technologies" - measured over centuries. Meanwhile all technologies (nearly) experience discontinuous advances, often near their start. See, steam engines and Watt, etc, etc, etc.

The rest of the argument seems to be grounded in an assumption that electronic neurons or sims of them won't ever be faster than meat. Really? Today's crude neural nets are already very useful here and there precisely because their speed means they scale and can repeat a task very, very frequently in a small amount of time.

Re: Likelihood of discontinuous progress around the development of AGI

#95

Post by yonkshi »

AndrewKemendo wrote: post #45
YeGoblynQueenne wrote: post #29

About AlphaZero particularly, a few things must be kept in mind. First, AlphaZero still makes use of a Monte Carlo Tree Search algorithm to search for good moves. MCTS is a powerful algorithm with a very limited scope: zero-sum, perfect information games. So for instance, it would be very difficult to see how to use MCTS-based AlphaZero in, e.g., training self-driving cars. Second, the AlphaZero architecture is preci…

So, AlphaZero doesn't mean we're closer to _general_ AI. Quite the contrary: it's a very specialised form of AI that will be very difficult to use in any different task than chess, shoggi or go. This is a very true statement and one that I think a lot of people who aren't in ML/DL, but are "worried" about AGI, miss. There is however a common thread with everyone in AI, that they tend to think of AGI as "One algorithm…

I think you are only thinking of supervised learning's capabilities, which I assume is the field of your AGI research? I'm working with reinforcement learning and DRL research, and RL was born to address the short coming of supervised learning. DeepMind is arguably the forerunner in RL right now, and AlphaZero is the crystalline of their RL research.

Yes, AlphaGo uses NN and other SL techniques, but the core is very much DQN based RL. No amount of SL can effectively play go and invent new moves. RL can already solve a large number of real world problems with a rather simple algorithm, from self driving cars to video games to NLP. RL can tackle all those problems with pretty much the same core algorithm. The question lies less in IF RL can solve more general AI problems, but rather HOW to solve it. From a high level view, we are having a lot of trouble with its convergence properties mathematically and its extreme sample inefficiency. This is the reason why Boston Dynamics doesn't use much RL, Waymo doesn't use much RL, simply because they can do much better with current techniques without going RL.

AlphaGo is still a major step forward regardless, because it's one of the biggest leap in RL we've taken in the recent years. It suddenly lets RL stably converge on solutions more than we could ever before. AlphaGo's contribution is more than just that it built a specialized Go bot, but rather a much more stable RL algorithm that lets us approximate non linear functions (majority of the real world applications are non linear). If I were to put my money, AI could very well be entering a new era with AlphaGo and their DQN.

Re: Likelihood of discontinuous progress around the development of AGI

#96

Post by AndrewKemendo »

sgt101 wrote: post #92
AndrewKemendo wrote: post #45

Earlier quoted context omitted.

So, AlphaZero doesn't mean we're closer to _general_ AI. Quite the contrary: it's a very specialised form of AI that will be very difficult to use in any different task than chess, shoggi or go. This is a very true statement and one that I think a lot of people who aren't in ML/DL, but are "worried" about AGI, miss. There is however a common thread with everyone in AI, that they tend to think of AGI as "One algorithm…

Well - SOAR.

What does this comment mean?

Re: Likelihood of discontinuous progress around the development of AGI

#97

Post by yonkshi »

YeGoblynQueenne wrote: post #29

About AlphaZero particularly, a few things must be kept in mind. First, AlphaZero still makes use of a Monte Carlo Tree Search algorithm to search for good moves. MCTS is a powerful algorithm with a very limited scope: zero-sum, perfect information games. So for instance, it would be very difficult to see how to use MCTS-based AlphaZero in, e.g., training self-driving cars. Second, the AlphaZero architecture is preci…

I don't agree with this sentiment, although I agree that AI is not nearly at the level of the hype that pop culture makes it out to be. AlphaZero is still a significant contribution to 'AGI' that shouldn't be buried.

It's true that AlphaZero's knowledge is unable to be generalized for other systems, but its biggest contribution is a _stable_ RL system that can solve problems that no other systems can. This is the first piece of the puzzle of more general AI. Generalization, I would consider as the second piece of the puzzle to more general AI. Generalization may be achieved with potential research in transfer learning, model based RL, symbolic network. But without a stable RL algorithm such as DQN as foundation, generalization has nothing to stand on.

Re: Likelihood of discontinuous progress around the development of AGI

#98

Post by s1dechnl »

yonkshi wrote: post #97
YeGoblynQueenne wrote: post #29

About AlphaZero particularly, a few things must be kept in mind. First, AlphaZero still makes use of a Monte Carlo Tree Search algorithm to search for good moves. MCTS is a powerful algorithm with a very limited scope: zero-sum, perfect information games. So for instance, it would be very difficult to see how to use MCTS-based AlphaZero in, e.g., training self-driving cars. Second, the AlphaZero architecture is preci…

I don't agree with this sentiment, although I agree that AI is not nearly at the level of the hype that pop culture makes it out to be. AlphaZero is still a significant contribution to 'AGI' that shouldn't be buried. It's true that AlphaZero's knowledge is unable to be generalized for other systems, but its biggest contribution is a _stable_ RL system that can solve problems that no other systems can. This is the fir…

Having not defined what intelligence is and having not declared the nature of General Intelligence, you're sure an aspect of clearly defined weak AI is a significant contribution to 'AGI'... Interesting.

> It's true that AlphaZero's knowledge is unable to be generalized for other systems

Interesting admission.

> This is the first piece of the puzzle of more general AI.

The first piece is generalized intelligence. Architecturally, it looks nothing like Alpha Zero. However, you feel :

> Generalization, I would consider as the second piece of the puzzle to more general AI.

How is that the second piece? It's the piece.

> Generalization may be achieved with potential research in transfer learning, model based RL, symbolic network. But without a stable RL algorithm such as DQN as foundation, generalization has nothing to stand on.

So you're of the belief that current approaches are compatible with and are the underpinning of Artificial General Intelligence while Hinton is convinced one needs to scrap it and start over. Sound advice is being ignored and there is a clearly entrenched decision to continue pushing along w/ iterating weak AI. I came here to test the waters.. The commentary and the K-value feedback I've received so far informs me quite profitably.

Re: Likelihood of discontinuous progress around the development of AGI

#99

Post by mannykannot »

whereismypwd wrote: post #70
mannykannot wrote: post #26

Earlier quoted context omitted.

I am not entirely convinced by the 'most problems are exponentially hard' argument, as problem-solving often has synergistic effects - e.g. consider calculus, developed in solving the problem of planetary orbits, but useful for so much more. On the other hand, trying to discern a general rule from historical data is complicated by the fact that the number of problem-solvers has been growing exponentially.

I think the better argument is that most "breakthroughs" are exponentially hard, or we wouldn't consider them breakthroughs!

'Exponential' means something more specific than 'big' or 'largest in class'.

Re: Likelihood of discontinuous progress around the development of AGI

#100

Post by sgt101 »

AndrewKemendo wrote: post #96
sgt101 wrote: post #92

Earlier quoted context omitted.

Well - SOAR.

What does this comment mean?

The point of soar was/is to provide a framework for a collection of problem solvers with diverse methods to work together in a single agent. This was developed because of the insight that humans use diverse strategies to negotiate the challenges of their environment. Or... Lots of weak Ai adding up fyi general Ai.
Post reply on HN