Humans Have Improved At Go Since AI Became The Best

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

Moderator: Ras

User avatar
towforce
Posts: 12695
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK
Full name: Graham Laight

Humans Have Improved At Go Since AI Became The Best

Post by towforce »

This is an actual scientific study:

https://www.newscientist.com/article/23 ... the-world/

This makes a lot of sense to me: when you play stronger opponents, you get better faster. You can learn and improve from playing weaker players, but it is necessary to put more time and effort into the learning process.

OTOH, if the gap between yourself and the opponent is too large, then learning becomes difficult again. When I've played IMs and GMs, my experience has been that everything was going fine, and then the sky fell down!

Meanwhile, for those of us who have "geekish" social skills, maybe ChatGPT can be our salvation! :lol:
Human chess is partly about tactics and strategy, but mostly about memory
daniel71
Posts: 146
Joined: Wed Aug 27, 2008 3:48 am

Re: Humans Have Improved At Go Since AI Became The Best

Post by daniel71 »

Did the Go Champion Return to Playing Go? Shame he retired from playing after he lost to a Go computer.
User avatar
towforce
Posts: 12695
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK
Full name: Graham Laight

Re: Humans Have Improved At Go Since AI Became The Best

Post by towforce »

daniel71 wrote: Tue Mar 14, 2023 1:41 pm Did the Go Champion Return to Playing Go? Shame he retired from playing after he lost to a Go computer.
Unfortunately not: https://en.wikipedia.org/wiki/Lee_Sedol
Human chess is partly about tactics and strategy, but mostly about memory
Uri Blass
Posts: 11119
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: Humans Have Improved At Go Since AI Became The Best

Post by Uri Blass »

towforce wrote: Tue Mar 14, 2023 9:47 am This is an actual scientific study:

https://www.newscientist.com/article/23 ... the-world/

This makes a lot of sense to me: when you play stronger opponents, you get better faster. You can learn and improve from playing weaker players, but it is necessary to put more time and effort into the learning process.

OTOH, if the gap between yourself and the opponent is too large, then learning becomes difficult again. When I've played IMs and GMs, my experience has been that everything was going fine, and then the sky fell down!

Meanwhile, for those of us who have "geekish" social skills, maybe ChatGPT can be our salvation! :lol:
Humans learn from engines not only by playing but also by analyzing games with them so the advantage of engines in improving playing strength of humans can be bigger than the advantage of only playing against stronger opponents.
User avatar
towforce
Posts: 12695
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK
Full name: Graham Laight

Re: Humans Have Improved At Go Since AI Became The Best

Post by towforce »

Uri Blass wrote: Wed Mar 15, 2023 11:36 am
towforce wrote: Tue Mar 14, 2023 9:47 am This is an actual scientific study:

https://www.newscientist.com/article/23 ... the-world/

This makes a lot of sense to me: when you play stronger opponents, you get better faster. You can learn and improve from playing weaker players, but it is necessary to put more time and effort into the learning process.

OTOH, if the gap between yourself and the opponent is too large, then learning becomes difficult again. When I've played IMs and GMs, my experience has been that everything was going fine, and then the sky fell down!

Meanwhile, for those of us who have "geekish" social skills, maybe ChatGPT can be our salvation! :lol:
Humans learn from engines not only by playing but also by analyzing games with them so the advantage of engines in improving playing strength of humans can be bigger than the advantage of only playing against stronger opponents.

Very good point. 8-)
Human chess is partly about tactics and strategy, but mostly about memory
User avatar
WinPooh
Posts: 276
Joined: Fri Mar 17, 2006 8:01 am
Location: Russia
Full name: Vladimir Medvedev

Re: Humans Have Improved At Go Since AI Became The Best

Post by WinPooh »

towforce wrote: Tue Mar 14, 2023 4:39 pm
daniel71 wrote: Tue Mar 14, 2023 1:41 pm Did the Go Champion Return to Playing Go? Shame he retired from playing after he lost to a Go computer.
Unfortunately not: https://en.wikipedia.org/wiki/Lee_Sedol
As far as I remember, after losing to AlphaGo this human champion made a streak of some dozens consecutive wins against strong humans (high dans level). His retirement was not immediately caused by losing the match.
User avatar
Ajedrecista
Posts: 2164
Joined: Wed Jul 13, 2011 9:04 pm
Location: Madrid, Spain.

Re: Humans have improved at Go since AI became the best?

Post by Ajedrecista »

Hello:
WinPooh wrote: Wed Mar 29, 2023 1:51 pm
towforce wrote: Tue Mar 14, 2023 4:39 pm
daniel71 wrote: Tue Mar 14, 2023 1:41 pm Did the Go Champion Return to Playing Go? Shame he retired from playing after he lost to a Go computer.
Unfortunately not: https://en.wikipedia.org/wiki/Lee_Sedol
As far as I remember, after losing to AlphaGo this human champion made a streak of some dozens consecutive wins against strong humans (high dans level). His retirement was not immediately caused by losing the match.
The match against AlphaGo was in March, 2016. According to the following site, Lee Sedol played a number of games after that match:

https://www.goratings.org/en/players/5.html

If I am not wrong:

Code: Select all

===================================================================================
Year         Games (B+W)        Wins (B+W)       Loses (B+W)       Avg. Opp. Rating
===================================================================================
2016 *        53 (26+27)        36 (18+18)        17 ( 8+ 9)        3448.23 (~#45)
-----------------------------------------------------------------------------------
2017          44 (21+23)        27 (15+12)        17 ( 6+11)        3448.07 (~#48)
-----------------------------------------------------------------------------------
2018          73 (37+36)        36 (19+17)        37 (18+19)        3484.15 (~#40)
-----------------------------------------------------------------------------------
2019          18 ( 7+11)         7 ( 2+ 5)        11 ( 5+ 6)        3417.78 (~#67)
-----------------------------------------------------------------------------------
TOTAL        188 (91+97)       106 (54+52)        82 (37+45)        3459.22 (#N/A)
-----------------------------------------------------------------------------------

* After the AlphaGo match.
B: Black.
W: White.
#X: number X of the rankings of that year (1st January).
N/A: not available.
I hope no typos.

Regards from Spain.

Ajedrecista.