If machines can beat us at games, does it make them more ?

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.

BrendanJNorman
Posts: 1490
Joined: Sun Feb 07, 2016 11:43 pm
Full name: Brendan J Norman

Re: If machines can beat us at games, does it make them more

Post by BrendanJNorman » Mon Jan 09, 2017 8:24 am

When a machine is capable of creating a human being without any outside help, we'll safely be able to say they're smarter than us.

While only we are capable of creating them, there can be no argument about who is more intelligent.

I'm not sure why this topic is debated so much.

It's like saying: "Who is more powerful; a slave or his master?"

And some people then arguing for the slave because he works harder than his master.

This is the current relationship between man and machine.

matthewlai
Posts: 793
Joined: Sun Aug 03, 2014 2:48 am
Location: London, UK
Contact:

Re: If machines can beat us at games, does it make them more

Post by matthewlai » Mon Jan 09, 2017 12:36 pm

Comparing intelligence like that doesn't make any sense.

Human brains and computers are optimized for very different things, and there are many things computers are "more intelligent" for, just like how there are many things humans are "more intelligent" for.

Computers have been beating humans at doing additions since the 1940s.

Humans are still much better at learning a wide variety of simple tasks quickly than computers today.
Disclosure: I work for DeepMind on the AlphaZero project, but everything I say here is personal opinion and does not reflect the views of DeepMind / Alphabet.

duncan
Posts: 10530
Joined: Mon Jul 07, 2008 8:50 pm

Re: If machines can beat us at games, does it make them more

Post by duncan » Mon Jan 09, 2017 9:47 pm

BrendanJNorman wrote:
When a machine is capable of creating a human being without any outside help, we'll safely be able to say they're smarter than us.
even turing did not ask for that

http://www.turing.org.uk/scrapbook/test.html

Turing's 'imitation game' is now usually called 'the Turing test' for intelligence.

syzygy
Posts: 4458
Joined: Tue Feb 28, 2012 10:56 pm

Re: If machines can beat us at games, does it make them more

Post by syzygy » Mon Jan 09, 2017 10:05 pm

What is white’s winning strategy?

Two kings are on either side of a pawn blockade. White has an opportunity to promote the pawn on F6 to a stronger piece. But that square is being guarded by the black king.

[D]8/3k4/4pPp1/3pP1P1/2pP4/2P5/8/4K3 w - - 0 1
An example problem where human intuition often triumphs over AI. David Ireland, Author provided

For white to win, the white king must move around the blockade via column A and force the black king away. Defeat for black is then inevitable.

Simple enough? Not entirely for a chess AI, which has more difficulty perceiving white’s advantage. This is because it would need to search to a depth of 20 ply to find white’s advantage. In this position, at 15 plies there are 10,142,484,904,590 possible positions (we tried computing to 20 plies but after one week of computation, we gave up).

Most computer chess programs won’t see the winning strategy. Instead, they will move the white king to the centre of the board which is the common strategy when there are only a few pieces on the board.

Human intuition is still a powerful force.
Sure!

Code: Select all

info depth 26 seldepth 42 multipv 1 score mate 17 nodes 815896 nps 3903808 tbhits 461 time 209 pv e1d2 d7e8 d2c2 e8d8 c2b2 d8d7 b2a3 d7e8 a3a4 e8f7 a4b5 f7f8 b5c5 f8e8 c5d6 e8f7 d6d7 f7f8 d7e6 f8e8 e6d6 e8f8 d6d7 f8g8 e5e6 g8h8 e6e7 h8h7 f6f7 h7g7 e7e8q
Or my private patzer engine:

Code: Select all

  0:00.066  27   2582098   +Mate17   1.Kd2 Kd8 2.Kc2 Kd7 3.Kb2 Ke8 4.Ka3 Kd7 5.Kb4 Ke8 6.Kb5 Kd7 7.Kc5 Ke8 8.Kd6 Kf7 9.Kd7 Kf8 10.Kxe6 Ke8 11.Kd6 Kf7 12.Kd7 Kf8 13.e6 Kg8 14.e7 Kh7 15.f7 Kg7 16.e8=Q Kh7 17.Qg8#

User avatar
hgm
Posts: 23795
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: If machines can beat us at games, does it make them more

Post by hgm » Mon Jan 09, 2017 10:10 pm

This position actually has become a classic by serving as an example for demonstrating the usefulness of transposition tables, which leave almost nothing of those zillions of nodes in a tree where only two Kings are walking.

Fairy-Max is a bit slower, but eventually sees a mate:

Code: Select all

 34	  #18 	329.5M	4:26.68	e1d1 d7d8 d1c2 d8e8 c2b2 e8f8 b2a3 f8f7 a3b4 f7e8 b4b5 e8f8 b5c6 f8e8 c6d6 e8f7 d6d7 f7f8 d7e6 f8e8 e6d6 e8f7 e5e6 f7f8 e6e7 f8e8 d6c6 e8f7 c6d7
33	  #19 	181.4M	2:12.90	e1d1 d7d8 d1c1 d8d7 c1b2 d7e8 b2a3 e8f7 a3b4 f7e8 b4b5 e8f7 b5c5 f7e8 c5d6 e8f7 d6d7 f7f8 d7e6 f8e8 e6d6 e8f8 d6d7 f8g8 d7e6
 32	+16.96 	18.6M  	0:14.23	e1d1 d7d8 d1c1 d8d7 c1b2 d7e8 b2a3 e8f7 a3a4 f7f8 a4b5 f8f7 b5c5 f7g8 c5d6 g8f7
 31	+10.41 	15.3M  	0:11.73	e1d1 d7d8 d1c1 d8d7 c1b2 d7e8 b2a3 e8f7 a3a4 f7f8 a4a5 f8e8 a5b4 e8f8 b4c5 f8g8 c5c6 g8f7 c6b6 f7g8
 30	+9.97 	10.0M  	0:07.32	e1d1 d7d8 d1c1 d8d7 c1b2 d7e8 b2a3 e8f7 a3b4 f7e8 b4b5 e8f7 b5c6 f7e8 c6d6 e8f7
 29	+4.95 	6.61M  	0:04.67	e1d1 d7d8 d1c1 d8d7 c1b2 d7e8 b2a3 e8f7 a3b4 f7e8 b4b5 e8f7 b5c6 f7e8 c6d6 e8f7 d6d7 f7f8 d7e6 f8e8 e6d5 e8f8 e5e6 f8e8 e6e7 e8f7 d5c4 f7e8 c4d5
 28	+4.88 	3.91M  	0:02.75	e1d1 d7d8 d1c1 d8d7 c1b2 d7e8 b2a3 e8f7 a3b4 f7e8 b4b5 e8f7 b5c6 f7e8 c6d6 e8f7 d6d7
 27	+4.90 	3.42M  	0:02.39	e1d1 d7d8 d1c1 d8d7 c1b2 d7e8 b2a3 e8f7 a3b4 f7e8 b4b5 e8f7 b5c6 f7e8 c6d6 e8f7
 26	+4.88 	2.01M  	0:01.43	e1d1 d7d8 d1c1 d8d7 c1b2 d7e8 b2a3 e8f7 a3b4 f7e8 b4b5 e8f7
 25	+4.29 	1.49M  	0:01.07	e1d1 d7d8 d1c1 d8d7 c1b2 d7d8 b2a3 d8e8 a3b2
 24	+4.29 	1.11M  	0:00.81	e1d1 d7d8 d1c1 d8d7 c1b2 d7d8 b2a3 d8e8 a3b2
 23	+3.58 	682373	0:00.51	e1d1 d7d8 d1c1 d8e8 c1b2 e8f7 b2a3 f7f8 a3b4 f8e8 b4b5 e8d8 b5c6 d8e8 c6d6 e8f7
 22	+3.52 	513721	0:00.39	e1d1 d7d8 d1c1 d8e8 c1b2 e8f7 b2a3 f7f8 a3b4 f8e8 b4b5 e8d8 b5c6 d8e8 c6d6 e8f7 d6d7 f7f8 d7e6 f8e8 e6d5 e8d7 d5c4
 21	+3.57 	436322	0:00.32	e1d1 d7d8 d1c1 d8e8 c1b2 e8f7 b2a3 f7f8 a3b4 f8e8 b4b5 e8d8 b5c6 d8e8 c6d6 e8f7 d6d7 f7f8 d7e6
 21	+1.95 	395424	0:00.29	e1e2 d7e8 e2d1 e8f7 d1c1 f7g8 c1b2 g8f8 b2a3 f8f7
 20	+1.48 	216722	0:00.17	e1e2 d7e8 e2e3 e8f7 e3f3 f7e8 f3f4 e8d7 f4g4 d7d8 g4g3 d8e8 g3f3 e8f8 f3g3
 19	+1.48 	151595	0:00.12	e1e2 d7e8 e2e3 e8f7 e3f3 f7e8 f3f4 e8d7 f4g3 d7d8 g3g4 d8d7 g4f4 d7d8 f4e3 d8d7 e3f2 d7d8 f2g2
 18	+1.48 	109829	0:00.09	e1e2 d7e8 e2e3 e8f7 e3f3 f7e8 f3f4 e8d7 f4g3 d7e8 g3f3 e8d7 f3f2 d7e8 f2e2
 17	+1.48 	80614  	0:00.07	e1e2 d7e8 e2e3 e8f7 e3f3 f7e8 f3f4 e8d7 f4e3 d7e8 e3f3 e8d7 f3f4 d7e8
 16	+1.48 	61328  	0:00.06	e1e2 d7e8 e2e3 e8f7 e3f3 f7e8 f3f4 e8d7 f4e3 d7e8 e3f3 e8f8 f3f4 f8e8 f4g4 e8f7
 15	+1.48 	41606  	0:00.04	e1e2 d7e8 e2e3 e8f7 e3f3 f7e8 f3f4 e8d7 f4e3 d7d8 e3f2 d8d7 f2g3 d7e8 g3f3
 14	+1.49 	27783  	0:00.03	e1e2 d7e8 e2e3 e8f7 e3f3 f7e8 f3f4 e8d7 f4e3 d7e8 e3f3 e8f8 f3f4 f8f7
 13	+1.48 	18274  	0:00.01	e1e2 d7e8 e2e3 e8f7 e3f3 f7e8 f3f4 e8d7 f4e3 d7e8 e3f3 e8f7 f3f4
 12	+1.48 	12939  	0:00.01	e1e2 d7e8 e2e3 e8f7 e3f3 f7e8 f3f4 e8d7 f4e3 d7e8 e3f3 e8d7
 11	+1.50 	8882    	0:00.01	e1e2 d7e8 e2e3 e8f7 e3f3 f7e8 f3f4 e8f8
 10	+1.49 	6355    	0:00.01	e1e2 d7e8 e2e3 e8f7 e3f4 f7g8 f4f3 g8f7 f3e3 f7f8
  9	+1.48 	4198    	0:00.01	e1e2 d7e8 e2e3 e8f7 e3f4 f7g8 f4f3 g8f7 f3e3
  8	+1.49 	2737    	0:00.01	e1e2 d7e8 e2e3 e8f7 e3f4 f7f8 f4f3
  7	+1.52 	1636    	0:00.01	e1e2 d7e8 e2e3 e8d7 e3f3 d7e8 f3f4
  6	+1.51 	1083    	0:00.01	e1e2 d7e8 e2f3 e8f7 f3f4
  5	+1.48 	630      	0:00.00	e1e2 d7e8 e2e3 e8f7 e3f4
  4	+1.48 	345      	0:00.00	e1e2 d7e8 e2e3 e8d7
  3	+1.52 	227      	0:00.00	e1e2 d7e8 e2e3
  3	+1.48 	166      	0:00.00	e1f1 d7e8 f1e2
  3	+0.05 	117      	0:00.00	f6f7 d7e7 f7f8 e7f8
  2	+1.98 	14        	0:00.00	f6f7 d7e7
  1	+1.99 	7          	0:00.00	f6f7
After 0.32 sec it already sees a way to make progress.

User avatar
Leto
Posts: 2028
Joined: Thu May 04, 2006 1:40 am
Location: Dune

Re: If machines can beat us at games, does it make them more

Post by Leto » Mon Jan 09, 2017 10:22 pm

Isn't it bizarre that the article claims their computer searched for a week and it didn't reach a depth of 20 ply on that problem? To me that sounds like something you'd expect from a computer from the 60's. Any computer today should have no problem reaching 20 ply in seconds. A recent dev version of Stockfish finds the mate in less than a second.

Dann Corbit
Posts: 10207
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: If machines can beat us at games, does it make them more

Post by Dann Corbit » Mon Jan 09, 2017 10:46 pm

Let's start with multiplication.
I think we can all safely agree that computers multiply faster than we do.
Even that Daniel guy from Great Britain can't do it as fast as a computer.

Let's go on to games:
Computers win.

Now computers can do other sorts of AI things too. They can fold towels.
http://news.berkeley.edu/2010/04/02/robot/
but it turns out that we are a bit better at it:
http://www.npr.org/sections/money/2015/ ... ing-towels

At some point in the next two decades, the computers will surpass us in full on brain horsepower.

I guess that if we keep going on the way we are, that computers would get the upper hand and take over.

The military seems to have overlooked Asimov's robot rules and are already building fully automated killing systems that can decide for themselves about targets.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

kinderchocolate
Posts: 411
Joined: Mon Nov 01, 2010 5:55 am
Full name: Ted Wong
Contact:

Re: If machines can beat us at games, does it make them more

Post by kinderchocolate » Mon Jan 09, 2017 11:43 pm

Leto wrote:Isn't it bizarre that the article claims their computer searched for a week and it didn't reach a depth of 20 ply on that problem? To me that sounds like something you'd expect from a computer from the 60's. Any computer today should have no problem reaching 20 ply in seconds. A recent dev version of Stockfish finds the mate in less than a second.
I believe the article author meant reaching 20 ply by brute-force. That is, compute all the possible moves.

He then compared the difficulty (> one week computation) with human intuition.

That was why I shared the article, because I think the author (who was a PHD in artificial intelligence) didn't understand what he was writing, something that is supposed to be his research area.

nionita
Posts: 161
Joined: Fri Oct 22, 2010 7:47 pm
Location: Austria

Re: If machines can beat us at games, does it make them more

Post by nionita » Tue Jan 10, 2017 12:35 am

This one was really good!

Barbarossa did not find the correct plan after 20 plys and I was disappointed.

Then I decided suddenly to inactivate null move search when side to move has no piece (I wanted to do it long before and never came to it). Et voila!

Up to ply 16 everything normal, but at ply 17 the new version doubbles the advantage for white on 576 cp and has already the best move! It took only abt 17k nodes and 16 milliseconds for this.

Post Reply