• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

AlphaGo AI vs Go champ Lee Sedol (Game over, humanity is done)

Status
Not open for further replies.
THE GUY ON THE LEFT is trying to make it interesting for regular people like me. He's not too bad :)

I'm a regular person have no clue on Go
still want less of Chris and more of the smarter guy. Because he actually has insight into who is ahead in real time.

edit: and he should stop playing with his freaking phone on camera while getting his question answered.
 
My only exposure to Go was the manga in Shonen Jump about a decade ago.. Hikaru no Go.

That manga got really intense and amazing even if 9 year old me had no idea how to play Go.
 
I'm a regular person have no clue on Go
still want less of Chris and more of the smarter guy. Because he actually has insight into who is ahead in real time.

edit: and he should stop playing with his freaking phone on camera while getting his question answered.

Should just let the pro do the narrating alone, the other commentator is no good
The guy on the left is an amateur 3d and therefore still stronger than a huge percentage of all Go players. He can't really talk up to Michael Redmond but he'd still rather easily demolish any one of us.

This bottom left corner is insane. I don't even know.
 
it's david vs goliath!

david = 3lbs of meat using 150 watts provided by its own portable power supply
goliath = a data center the size of a football field sucking down half a megawatt of power.
 
it's david vs goliath!

david = 3lbs of meat using 150 watts provided by its own portable power supply
goliath = a data center the size of a football field sucking down half a megawatt of power.

But, that's what people said about computers 60 years ago. In ten years, you will have that processing power in your pocket.
 
For a hot minute we reached the limit of vacuum tubes, who's to say some alternate technology to the silicon transistor won't pop up?

Who's to say something will? That's the real problem.

Also, the limits of vacuum tubes are limits of engineering. The limit of silicon transistors is one of physics. That is a whole order of magnitude more difficult to crack.
 
Who's to say something will? That's the real problem.

Also, the limits of vacuum tubes are limits of engineering. The limit of silicon transistors is one of physics. That is a whole order of magnitude more difficult to crack.

There is already research going into silicon alternatives which get around the heat problem. With an industry in the billions that hinges on finding a solution to this, my bet is that humanity will.
 
I really should get my board out, and start learning to play this game properly...

Can't even beat GNUGo on the lowest difficulty level lol!
 
This expert is an amazing comentator. Thanks to him I feel like I know how to play even though I only learned this game exists a week ago
 
Man, this would be so hard to compute. So many combinations and permutations you'd have to search (and heuristics to run).
 
There is already research going into silicon alternatives which get around the heat problem. With an industry in the billions that hinges on finding a solution to this, my bet is that humanity will.

Heat isn't even the main issue, its quantum tunneling. And yeh that's a fundamental physical effect that happens as we try build transistors on the quantum scale. Quantum computing is the future of computational power.
 
For a hot minute we reached the limit of vacuum tubes, who's to say some alternate technology to the silicon transistor won't pop up?

I think the problem is we're trying to achieve acceptable artificial intelligence via a still flawed route. Throwing more transistors at it doesn't help. It isn't like a turing-candidate AI works acceptably if we give it weeks instead of seconds to think about its responses.
 
Man, this would be so hard to compute. So many combinations and permutations you'd have to search (and heuristics to run).
Imagine how humans feel when they play this game.

Speaking of which, if anyone wants to learn/play some go, I highly recommend it and can teach the basics and stuff.
 
AIs have been obliterating humans at complex games since the 90s. Go is just a tougher one.

Yes, but from what I understand, Go is a magnitudes more difficult game than say chess, and I remember 10 or 15 years ago people wondered if a computer could ever play Go competitively because moves can't be determined by the kind of normal decision tree that would be used for a chess kind of game.
 
I really hope it beats the human

Traitor!

No seriously though I hope AlphaGo wins at least one match. I don't think it's going to win this one though. Not that I have any knowledge of Go but listening to two streams' commentary I get the feeling Lee is at a strong advantage.
 
As someone who knows nothing about how to play Go, I hate how these announcers are going into deep conversations about how the end game would work and missing the actual plays. I wish they would focus more on what's actually happening, but admittedly it's probably hard to fill the time just doing that.
 
I'm assuming it takes the clock into account as well?

I was wondering that myself. Does it even know the clock exists? Because one would assume it would be programmed to cut down on how deep it computes as the time gets lower

Actually, what happens once the clock ends? Does the game just end for both people, or does the other person get some free reign to rack up points?
 
As someone who knows nothing about how to play Go, I hate how these announcers are going into deep conversations about how the end game would work and missing the actual plays. I wish they would focus more on what's actually happening, but admittedly it's probably hard to fill the time just doing that.
They're reading ahead to tell you what the implications of the moves are, since they're not clear to anyone except the highest level of professional.

Traitor!

No seriously though I hope AlphaGo wins at least one match. I don't think it's going to win this one though. Not that I have any knowledge of Go but listening to two streams' commentary I get the feeling Lee is at a strong advantage.
I think the game's a bit unpredictable right now, but again, I'm no expert. As far as I can tell, Lee Sedol is ahead but not by much.
 
I was wondering that myself. Does it even know the clock exists? Because one would assume it would be programmed to cut down on how deep it computes as the time gets lower

...actually, what happens once the clock ends; does the game just end for both people, or does the other person get some free reign to rack up points?
The sack of meat is taking a while too it seems. And we know this thing is going to be at least 2x, understating this, soon.
 
I was wondering that myself. Does it even know the clock exists? Because one would assume it would be programmed to cut down on how deep it computes as the time gets lower

Actually, what happens once the clock ends? Does the game just end for both people, or does the other person get some free reign to rack up points?

once you used the allocated time, you have to put your next move within 60 seconds i think... i forgot a lot of go rules
 
I was wondering that myself. Does it even know the clock exists? Because one would assume it would be programmed to cut down on how deep it computes as the time gets lower

Actually, what happens once the clock ends? Does the game just end for both people, or does the other person get some free reign to rack up points?

3 60 second byo yomi periods. It keeps resetting as long as you move in inder 60 seconds

If you go over 60 seconds more than 3 times you forfeit
 
once you used the allocated time, you have to put your next move within 60 seconds

So the times listed are a bank of time you can pull out of when making your moves, but if you run out of banked time, you are just forced to respond within 60 seconds? It doesn't force the game to end?

3 60 second byo yomi periods. It keeps resetting as long as you move in inder 60 seconds

If you go over 60 seconds more than 3 times you forfeit
... and then if you go over 60 seconds 3 times you forfeit?
 
funny now how the human play with experience and the AI need to evaluate every move

What do you mean?

The AI also plays from experience, since AlphaGo is heavily based on machine learning methods (key word being learning).
 
How long do they typically think before doing a move? How long do games take? What is this? How did that?
 
Status
Not open for further replies.
Top Bottom