Watching this, I love how the commentator mimics then move and then moves it immediately, thinking it has to be wrong, then wonders if it is a mistake and then towards the end of the clip appears to see the benefits of the move.
+Adam O'Halloran Yeah, a 9 Dan Professional "appears" to see its benefits.... Or maybe the move was so unusual, it just took him a moment to read out the possibilities of the move?
Michael: "...It changes the value of an area when you have a strong group like this, because Black doesn't have any point to approach it, because it's so strong." AlphaGo: "I beg to differ."
The moment the world realised machines could be creative, their reaction says it all haha. Interesting times we're going into that's for sure. Lets hope we don't screw it up.
I'm glad to see there is a video of it. I reference move 37 and use it as a metaphor in my upcoming novel, Hot Nights and Cold Wars: None of the Above 2. I highly recommend watching the full documentary. It's definitely time well spent. th-cam.com/video/WXuK6gekU1Y/w-d-xo.html In the documentary, Lee Sedol left before move 37. He returned from a break to that move. He normally made his next move in 1-2 minutes but took 12 minutes to respond to move 37.
2:18 "... black probably is thinking..." Did he say thinking? Alpha is thinking? lol. That settles it then. Artificial Intelligence has finally become reality.
It was a revolutionary move which has been very hard to digest. Here is the view from a year later: "AlphaGo played a handful of highly inventive winning moves, several of which - including move 37 in game two - were so surprising they overturned hundreds of years of received wisdom, and have since been examined extensively by players of all levels. In the course of winning, AlphaGo somehow taught the world completely new knowledge about perhaps the most studied and contemplated game in history." (deepmind.com/research/alphago/)
It's on the 5th line, so the traditional thinking is that W gets too much territory by just pushing on the 4th line. Shoulder hits are common but they are usually lower, on the 3rd or 4th line.
Nah, Sai from Hikaru no Go never got to face so many different pro in the modern era before and especially the reigning champion. So unless he could beat those people easily as Alphago did, he would be crush just like the rest by Alphago.
Sai was as strong as Shusaku, pretty strong even by today's standards but not the strongest. He wouldn't do better against AlphaGo than elite pros of today.
Fascinating! Neural networks probing neural networks, combining with similarly trained neural networks to repeat the same process... How soon before it breaks the rules in such a way as to emphatically signal it knows it is playing and it knows it's playing "us"...?
In your second comment, you use both "machine" and "humans" before identifying either with the pronoun "one" - which makes it very ambiguous. Is it the human thinking outside the box, rendering the machine's system useless? Or is it the machine thinking outside the box, rendering the human's system useless? Or, humorously, you could easily just say, "yes", lol. Cheers.
It's incredibly impressive. I would not say this really moved us much closer to artificial intelligence though. Essentially all it does is contain a massive filling cabinet of go games far larger than any human or group of humans could ever remember and reads through all the options and odds before each move. Its brilliant and simple at the same time, just like go.
This was a simulated neural network, that was trained by reviewing thousands and thousands of human go games followers by playing its self though thousands of iterations... in short to LEARN how to play go. Imaging having thousands of children raised to learn and do nothing other than play go, you then have these children compete against each other and kill the losers until one final child is left, that being your champion go player.
No, it doesn't use the brute force method there are too many combinations. When it was first made it was dumb, they let it run and practice with itself and learn and it got this good pretty quick.
"and it was in this position that lee sedol resigned" -agadmator
we all Agadmator fans, woahh!
Chess got me hooked on these types of games! If only there was an agadmator equivalent of go
This is an important moment in history
The biggest in computing history
@@JonathanXLindqviust ok hold your horses
Watching this, I love how the commentator mimics then move and then moves it immediately, thinking it has to be wrong, then wonders if it is a mistake and then towards the end of the clip appears to see the benefits of the move.
+Adam O'Halloran Yeah, a 9 Dan Professional "appears" to see its benefits.... Or maybe the move was so unusual, it just took him a moment to read out the possibilities of the move?
Yes, that is the point I was making...
I love his triple take
lol I counted five 😂
Lee was already gone for a smoke break when AlphaGo dropped this bomb. You'll see his confused reaction when he returns and sees Move 37 on the board.
Michael: "...It changes the value of an area when you have a strong group like this, because Black doesn't have any point to approach it, because it's so strong."
AlphaGo: "I beg to differ."
LOL. Thank you for drawing attention back to the commentary just as AlphaGo was actually making the move. Great contrast.
You don't.understand the move. Alpha go was forcing W to take the low valued area while building the center.
@@johntuttle9544
"and in that moment, the world was changed..."
well at least the humans noticed that it had changed.
Why does this have 60K views and not 100 million, like it should? This is a dramatic moment, one of the biggest breakthroughs in history. Goosebumps.
Because there are other videos showing the same
Also not so really
explain
AlphaGo finds the legendary Divine Move.
The moment the world realised machines could be creative, their reaction says it all haha. Interesting times we're going into that's for sure. Lets hope we don't screw it up.
on game 5 they explain move 37 , i think it was around 16 mins but idk
Outstanding Move
Skynet begins to learn at a geometric rate. It becomes self-aware at 2:14 PM, Korean Standard Time, March 16th. In a panic, they try to pull the plug.
Move 37 comes up
*Alpha Go:* I am about to end this mans whole career
I'm glad to see there is a video of it. I reference move 37 and use it as a metaphor in my upcoming novel, Hot Nights and Cold Wars: None of the Above 2. I highly recommend watching the full documentary. It's definitely time well spent. th-cam.com/video/WXuK6gekU1Y/w-d-xo.html
In the documentary, Lee Sedol left before move 37. He returned from a break to that move. He normally made his next move in 1-2 minutes but took 12 minutes to respond to move 37.
Who on earth calls it a "clicko"? Come on, Chris.
Anyone here from Life 3.0
alphago is a beast, just beat the best player in the world by 0.5 points
2:18 "... black probably is thinking..." Did he say thinking? Alpha is thinking? lol. That settles it then. Artificial Intelligence has finally become reality.
It's a neural network, not as complex as our brain, but it is thinking.
go check sophia hanson
so was it a good move or a mistake?
It was a revolutionary move which has been very hard to digest. Here is the view from a year later: "AlphaGo played a handful of highly inventive winning moves, several of which - including move 37 in game two - were so surprising they overturned hundreds of years of received wisdom, and have since been examined extensively by players of all levels. In the course of winning, AlphaGo somehow taught the world completely new knowledge about perhaps the most studied and contemplated game in history." (deepmind.com/research/alphago/)
It was SO good, humans could not understand it at first. We now know, many years later, it was the first hint that AI could become super human.
I thought Lee left before the move, not after.
But why is the move unusual ? Why doesn't people normally use that move ?
DeepMind CEO Demis Hassabis gives an excellent overview of move 37 in this talk, starting at 27:30: th-cam.com/video/HmvEzCqC16g/w-d-xo.html
Oh, thanks a lot.
It's on the 5th line, so the traditional thinking is that W gets too much territory by just pushing on the 4th line. Shoulder hits are common but they are usually lower, on the 3rd or 4th line.
Updated link to DeepMind CEO Demis Hassabis talking about move 37: (at 27:30) - th-cam.com/video/Y99kD9yqdlc/w-d-xo.htmlm20s
@Mmmmmh no sound on that video??
Who is Lee in the video?
Lee Sedol
Still don't know how to play go...
He left the room *Before* that move.
Mi no entender... :D
he rage quited xD
I don't think he quit. I think he just left the room to gather his composure.
Nope, he left the room before the move was played by Alphago
Hurray AlphaGo !
maybe the first step in the singularity ?
Sai from Hikaru no Go would have crushed AlphaGo Fan, Lee or Zero easily. :D
Nah, Sai from Hikaru no Go never got to face so many different pro in the modern era before and especially the reigning champion. So unless he could beat those people easily as Alphago did, he would be crush just like the rest by Alphago.
Sai was as strong as Shusaku, pretty strong even by today's standards but not the strongest. He wouldn't do better against AlphaGo than elite pros of today.
Ah.
When God enters the game!
Fascinating! Neural networks probing neural networks, combining with similarly trained neural networks to repeat the same process... How soon before it breaks the rules in such a way as to emphatically signal it knows it is playing and it knows it's playing "us"...?
In your second comment, you use both "machine" and "humans" before identifying either with the pronoun "one" - which makes it very ambiguous. Is it the human thinking outside the box, rendering the machine's system useless? Or is it the machine thinking outside the box, rendering the human's system useless? Or, humorously, you could easily just say, "yes", lol.
Cheers.
And somehow he still lost the 5th match.
Life 3.0
mejor vuelvo a mi cueva :'v
I don't understand at all, but skynet is real
Wtf is this
It's incredibly impressive. I would not say this really moved us much closer to artificial intelligence though. Essentially all it does is contain a massive filling cabinet of go games far larger than any human or group of humans could ever remember and reads through all the options and odds before each move. Its brilliant and simple at the same time, just like go.
No, it specifically does not retain memory of game moves. It is a neural network.
This was a simulated neural network, that was trained by reviewing thousands and thousands of human go games followers by playing its self though thousands of iterations... in short to LEARN how to play go.
Imaging having thousands of children raised to learn and do nothing other than play go, you then have these children compete against each other and kill the losers until one final child is left, that being your champion go player.
You don't need to kill them.
this just shows that we gained the catalyst for ai to reach higher than human thinking
No, it doesn't use the brute force method there are too many combinations. When it was first made it was dumb, they let it run and practice with itself and learn and it got this good pretty quick.