I'd like to see him take on AI with his own AI programs next.
Go master Lee Se-dol says he quits, unable to win over AI Go players
461–470 of 554 posts
Re: Go master Lee Se-dol says he quits, unable to win over AI Go players
#462Earlier quoted context omitted.
Go still isn't solved (neither is chess), we just have a machine good at knowing which parts of the search space are worth checking.
I think achieving superiority over humans is practically solving the problem though. Solving chess or go by going through a complete search space seems more like a hardware/computational goal than a practical ml/ai goal.
We've just created a tool people can use to play Go better than a person without the tool. Until something emerges from the rain forest, or comes down from space that can throw down and win I'd say Humans are still the best Go players in the known universe.
Re: Go master Lee Se-dol says he quits, unable to win over AI Go players
#463By definition, if a system wins even 10% of the time against you and gives you a headache, while it's doing it in a breeze, you're going to give up. It's unsustainable.
Re: Go master Lee Se-dol says he quits, unable to win over AI Go players
#464Is there any attempt to normalize the time available to the human and machine players? Seem inherently unfair to give the computer player the same clock time to make a move.
Re: Go master Lee Se-dol says he quits, unable to win over AI Go players
#465Isn't saying AI is the future of Go like saying cars are the future of sprinting? I mean who cares if machines are faster/better/smarter than us at any particular task? That is true in millions of ways where humans still enjoy competing against each other. Maybe what's needed is just a perspective change where we stop thinking of Go as being against any other agent and make it, like every other human competition, aga…
It's bizarrely "hip" or "woke" or whatever right now to anthropomorphize "AI". "AI" is light years away from being anything more than a tool. AI doesn't "play Go" in my view as much as people play go, via AI. Is it really shocking we've built a tool that we can use to play Go better than somebody without said tool?
If Lee Sedol says "Okay AlphaGo, let's play!" and sits down at a board what happens? Nothing happens! AlphaGo has no agency. AlphaGo is an extension of human agency. AlphaGo isn't better than humans at Go. Humans with AlphaGo are better at Go than humans without AlphaGo.
Re: Go master Lee Se-dol says he quits, unable to win over AI Go players
#466Isn't saying AI is the future of Go like saying cars are the future of sprinting? I mean who cares if machines are faster/better/smarter than us at any particular task? That is true in millions of ways where humans still enjoy competing against each other. Maybe what's needed is just a perspective change where we stop thinking of Go as being against any other agent and make it, like every other human competition, aga…
I don't think this is about GO, but about humans becoming useless faster every day against machines in virtually everything and how that is going to affect society and us as humans.
Re: Go master Lee Se-dol says he quits, unable to win over AI Go players
#467I sympathize. It used to be that you might be able to believe that there is some kind of art behind go, some sort of abstract beauty to it, and that the pursuit of this beauty is the path to being good at go ... But the defeat of the tactics born from this mindset by MCTS at least for now lay bare the fact that the path to being good at go is actually to probabilistically sample the phase space of the game and perfor…
>> But the defeat of the tactics born from this mindset by MCTS at least for now lay bare the fact that the path to being good at go is actually to probabilistically sample the phase space of the game and perform impossibly mindless statistics on the game outcome an enormous number of times ... This is a rationalist opinion that is attractive because it appears unbiased, unwilling to make any concessions to human nat…
> In particular, human players do none of the extremely tedious, extremely computationally intensive maths that computer players have to do.
This sounds like a Chinese room argument.
Performing computationally expensive maths doesn’t make the computer intelligent.
But that says nothing about the intelligence of the maths itself.
Re: Go master Lee Se-dol says he quits, unable to win over AI Go players
#468Rubber meet road.
Re: Go master Lee Se-dol says he quits, unable to win over AI Go players
#469Earlier quoted context omitted.
Well the energy spent creating you across the last billion years was also significant!
Then the energy spent creating the machine is always higher
If you're going to count that though then the difference in energy between the human and machine become relatively tiny and the comparison just becomes a comparison of strength again. :)
I think runtime energy is the most interesting.
Re: Go master Lee Se-dol says he quits, unable to win over AI Go players
#470Earlier quoted context omitted.
Is this true? From what I've read about Alpha Zero/etc, in both Go and chess, it's going interesting move sequences that people hadn't considered viable before. That certainly seems like an interesting thing to learn. Also, sure, you can dismiss it all as statistics. But how sure are you that what's happening with humans isn't in some form? I'd also say that MCTS is something people kind of do too in games: look a fe…
No, it isn't. The top humans are much better at Go than they were four years ago, largely due to learning from the new engines. If it were all just about sampling the phase space zillions of times, this would not be the case.
This is interesting to me ... how exactly is this assessed ...?
Do people keep around versions of “alpha go year 2017” and play against it in order to measure human improvement over time?
If the basis for observing improvement has become “I can beat old versions of the ai more reliably than I used to be able to” or if I have learned to beat players who have not studied alpha zero I suppose that’s a form of usefully learning “about go” by analyzing the games played by alpha zero ...
I wonder if we might ever arrive at a point where human vs fixed-year x ai performance at go pretty much stops increasing over time ...?