View source for Software:AlphaGo

{{short description|Artificial intelligence that plays Go}}
{{Use Oxford spelling|date=September 2016}}
{{Infobox software
| name = AlphaGo
| logo = Alphago_logo_Reversed.svg
| developer = [[Company:DeepMind|DeepMind]]
| genre = [[Software:Computer Go|Computer Go]] [[Software:Go software|software]]
| website = [https://www.deepmind.com/research/highlighted-research/alphago deepmind.com/research/highlighted-research/alphago]
}}
{{Artificial intelligence}}
'''AlphaGo''' is a [[Computer program|computer program]] that plays the [[Engineering:Board game|board game]] [[Software:Go (game)|Go]].<ref>{{cite news|url=https://www.bbc.com/news/technology-35785875|title=Artificial intelligence: Google's AlphaGo beats Go master Lee Se-dol|work=BBC News|date=12 March 2016|access-date=17 March 2016|archive-date=26 August 2016|archive-url=https://web.archive.org/web/20160826103910/http://www.bbc.com/news/technology-35785875|url-status=live}}</ref> It was developed by the London-based [[Company:DeepMind|DeepMind]] Technologies,<ref>{{cite web |url=https://deepmind.com/research/case-studies/alphago-the-story-so-far |title=DeepMind AlphaGO |work=DeepMind Artificial Intelligence AlphaGo |access-date=16 September 2019 |archive-date=14 September 2019 |archive-url=https://web.archive.org/web/20190914102339/https://deepmind.com/research/case-studies/alphago-the-story-so-far |url-status=live }}</ref> an acquired subsidiary of [[Company:Google|Google]] (now Alphabet Inc.). Subsequent versions of AlphaGo became increasingly powerful, including a version that competed under the name Master.<ref>{{cite web |title=AlphaGo {{!}} DeepMind |url=https://deepmind.com/research/alphago/ |website=DeepMind |access-date=28 May 2017 |archive-date=28 May 2017 |archive-url=https://web.archive.org/web/20170528120015/https://deepmind.com/research/alphago/ |url-status=live }}</ref> After retiring from competitive play, AlphaGo Master was succeeded by an even more powerful version known as [[AlphaGo Zero]], which was completely [[Self-play (reinforcement learning technique)|self-taught]] without learning from human games. AlphaGo Zero was then generalized into a program known as [[AlphaZero]], which played additional games, including chess and shogi.  AlphaZero has in turn been succeeded by a program known as [[MuZero]] which learns without being taught the rules.

AlphaGo and its successors use a [[Monte Carlo tree search]] algorithm to find its moves based on knowledge previously acquired by [[Machine learning|machine learning]], specifically by an [[Artificial neural network|artificial neural network]] (a [[Deep learning|deep learning]] method) by extensive training, both from human and computer play.<ref name="DeepMindnature2016"/> A neural network is trained to identify the best moves and the winning percentages of these moves. This neural network improves the strength of the tree search, resulting in stronger move selection in the next iteration.

In October 2015, in a match against Fan Hui, the original AlphaGo became the first [[Software:Computer Go|computer Go]] program to beat a human professional Go player without handicap on a full-sized 19×19 board.<ref name="googlego">{{cite web|url=http://googleresearch.blogspot.com/2016/01/alphago-mastering-ancient-game-of-go.html|title=Research Blog: AlphaGo: Mastering the ancient game of Go with Machine Learning|date=27 January 2016|work=Google Research Blog|access-date=28 January 2016|archive-date=30 January 2016|archive-url=https://web.archive.org/web/20160130003400/http://googleresearch.blogspot.com/2016/01/alphago-mastering-ancient-game-of-go.html|url-status=live}}</ref><ref name="bbcgo" /> In March 2016, it beat Lee Sedol in a five-game match, the first time a computer Go program has beaten a [[Software:Go ranks and ratings|9-dan]] professional without handicap.<ref name="leesedolwin">{{cite web |url=https://www.youtube.com/watch?v=vFr3K2DORc8&t=1h57m |title=Match 1 – Google DeepMind Challenge Match: Lee Sedol vs AlphaGo |website=[[Company:YouTube|YouTube]] |date=8 March 2016 |access-date=9 March 2016 |archive-date=29 March 2017 |archive-url=https://web.archive.org/web/20170329005648/https://www.youtube.com/watch?v=vFr3K2DORc8&t=1h57m |url-status=live }}</ref> Although it lost to Lee Sedol in the fourth game, Lee resigned in the final game, giving a final score of 4 games to 1 in favour of AlphaGo. In recognition of the victory, AlphaGo was awarded an honorary 9-dan by the Korea Baduk Association.<ref>{{cite news|url=http://www.straitstimes.com/asia/east-asia/googles-alphago-gets-divine-go-ranking|title=Google's AlphaGo gets 'divine' Go ranking|newspaper=The Straits Times|publisher=straitstimes.com|date=15 March 2016|access-date=9 December 2017|archive-date=7 October 2016|archive-url=https://web.archive.org/web/20161007023450/http://www.straitstimes.com/asia/east-asia/googles-alphago-gets-divine-go-ranking|url-status=live}}</ref> The lead up and the challenge match with Lee Sedol were documented in a documentary film also titled ''AlphaGo'',<ref name="alphagomovie">{{Cite web|url=https://www.alphagomovie.com/|title=AlphaGo Movie|website=AlphaGo Movie|access-date=14 October 2017|archive-date=3 January 2018|archive-url=https://web.archive.org/web/20180103003004/https://www.alphagomovie.com/|url-status=live}}</ref> directed by Greg Kohs. The win by AlphaGo was chosen by ''Science'' as one of the Breakthrough of the Year runners-up on 22 December 2016.<ref>{{cite journal|url=https://www.science.org/content/article/ai-protein-folding-our-breakthrough-runners|title=From AI to protein folding: Our Breakthrough runners-up|journal=Science|date=22 December 2016|access-date=29 December 2016|archive-date=17 June 2022|archive-url=https://web.archive.org/web/20220617185537/https://www.science.org/content/article/ai-protein-folding-our-breakthrough-runners|url-status=live}}</ref>

At the 2017 [[Software:Future of Go Summit|Future of Go Summit]], the Master version of AlphaGo beat Ke Jie, the number one ranked player in the world at the time, in a three-game match, after which AlphaGo was awarded professional 9-dan by the Chinese Weiqi Association.<ref name="AlphaGo 9-dan">{{cite web|url=http://sports.sohu.com/20170527/n494734669.shtml|title=中国围棋协会授予AlphaGo职业九段 并颁发证书|language=zh|publisher=Sohu.com|date=27 May 2017|access-date=9 December 2017|archive-date=3 June 2017|archive-url=https://web.archive.org/web/20170603045548/http://sports.sohu.com/20170527/n494734669.shtml|url-status=live}}</ref>

After the match between AlphaGo and Ke Jie, DeepMind retired AlphaGo, while continuing AI research in other areas.<ref name="AlphaGo's Designers Explore New AI">{{cite magazine|url=https://www.wired.com/2017/05/win-china-alphagos-designers-explore-new-ai/|title=After Win in China, AlphaGo's Designers Explore New AI|magazine=Wired|date=2017-05-27|last1=Metz|first1=Cade}}</ref> The self-taught AlphaGo Zero achieved a 100–0 victory against the early competitive version of AlphaGo, and its successor [[AlphaZero]] was perceived as the world's top player in Go by the end of the 2010s.<ref>{{cite web |url=https://www.chess.com/news/view/updated-alphazero-crushes-stockfish-in-new-1-000-game-match |title=AlphaZero Crushes Stockfish In New 1,000-Game Match |date=17 April 2019 |access-date=22 April 2021 |archive-date=12 November 2020 |archive-url=https://web.archive.org/web/20201112023708/https://www.chess.com/news/view/updated-alphazero-crushes-stockfish-in-new-1-000-game-match |url-status=live }}</ref><ref>{{cite journal |url=https://www.science.org/doi/10.1126/science.aar6404 |title=A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play |date=7 December 2018 |doi=10.1126/science.aar6404 |last1=Silver |first1=David |last2=Hubert |first2=Thomas |last3=Schrittwieser |first3=Julian |last4=Antonoglou |first4=Ioannis |last5=Lai |first5=Matthew |last6=Guez |first6=Arthur |last7=Lanctot |first7=Marc |last8=Sifre |first8=Laurent |last9=Kumaran |first9=Dharshan |last10=Graepel |first10=Thore |last11=Lillicrap |first11=Timothy |last12=Simonyan |first12=Karen |last13=Hassabis |first13=Demis |journal=Science |volume=362 |issue=6419 |pages=1140–1144 |pmid=30523106 |bibcode=2018Sci...362.1140S |s2cid=54457125 |access-date=30 June 2022 |archive-date=3 June 2022 |archive-url=https://web.archive.org/web/20220603194830/https://www.science.org/doi/10.1126/science.aar6404 |url-status=live |doi-access=free }}</ref>

==History==
Go is considered much more difficult for computers to win than other games such as chess, because its strategic and aesthetic nature makes it hard to directly construct an evaluation function, and its much larger [[Branching factor|branching factor]] makes it prohibitively difficult to use traditional AI methods such as [[Alpha–beta pruning|alpha–beta pruning]], [[Tree traversal|tree traversal]] and [[Heuristic|heuristic]] search.<ref name="googlego" /><ref>{{citation|title=Temporal Difference Learning of Position Evaluation in the Game of Go|first1=Nicol N.|last1=Schraudolph|first2=Peter Dayan|last2=Terrence|first3=J.|last3=Sejnowski|url=http://www.variational-bayes.org/~dayan/papers/sds94.pdf|access-date=31 January 2016|archive-date=28 March 2017|archive-url=https://web.archive.org/web/20170328153822/http://www.variational-bayes.org/~dayan/papers/sds94.pdf|url-status=live}}</ref>

Almost two decades after [[Company:IBM|IBM]]'s computer Deep Blue beat world chess champion Garry Kasparov in the 1997 match, the strongest Go programs using [[Artificial intelligence|artificial intelligence]] techniques only reached about [[Software:Go professional#Pro and amateur dan|amateur 5-dan]] level,<ref name="DeepMindnature2016"/> and still could not beat a professional Go player without a handicap.<ref name="googlego" /><ref name="bbcgo" /><ref name="CNN0128">{{cite web |url=https://money.cnn.com/2016/01/28/technology/google-computer-program-beats-human-at-go/index.html |title=Computer scores big win against humans in ancient game of Go |publisher=CNN |date=28 January 2016 |access-date=28 January 2016 |archive-date=31 January 2016 |archive-url=https://web.archive.org/web/20160131060936/http://money.cnn.com/2016/01/28/technology/google-computer-program-beats-human-at-go/index.html |url-status=live }}</ref> In 2012, the software program [[Software:Zen|Zen]], running on a four PC cluster, beat Masaki Takemiya ([[Software:Go professional|9p]]) twice at five- and four-stone handicaps.<ref>{{cite web|url=https://gogameguru.com/zen-computer-go-program-beats-takemiya-masaki-4-stones/|title=Zen computer Go program beats Takemiya Masaki with just 4 stones!|work=Go Game Guru|access-date=28 January 2016|archive-url=https://web.archive.org/web/20160201162313/https://gogameguru.com/zen-computer-go-program-beats-takemiya-masaki-4-stones/|archive-date=1 February 2016|url-status=dead|df=dmy-all}}</ref> In 2013, [[Software:Crazy Stone|Crazy Stone]] beat Yoshio Ishida (9p) at a four-stone handicap.<ref>{{cite web|title=「アマ六段の力。天才かも」囲碁棋士、コンピューターに敗れる 初の公式戦 |url=http://sankei.jp.msn.com/life/news/130320/igo13032020420000-n1.htm |publisher=MSN Sankei News |access-date=27 March 2013 |url-status=dead |archive-url=https://web.archive.org/web/20130324221549/http://sankei.jp.msn.com/life/news/130320/igo13032020420000-n1.htm |archive-date=24 March 2013 }}</ref>

According to DeepMind's [[Biography:David Silver (programmer)|David Silver]], the AlphaGo research project was formed around 2014 to test how well a neural network using [[Deep learning|deep learning]] can compete at Go.<ref name=pcworld_unusual>{{cite news|author1=John Riberio|title=AlphaGo's unusual moves prove its AI prowess, experts say|url=http://www.pcworld.com/article/3043668/analytics/alphagos-unusual-moves-prove-its-ai-prowess-experts-say.html|access-date=18 March 2016|work=PC World|date=14 March 2016|archive-date=17 July 2016|archive-url=https://web.archive.org/web/20160717202329/http://www.pcworld.com/article/3043668/analytics/alphagos-unusual-moves-prove-its-ai-prowess-experts-say.html|url-status=live}}</ref> AlphaGo represents a significant improvement over previous Go programs. In 500 games against other available Go programs, including Crazy Stone and Zen, AlphaGo running on a single computer won all but one.<ref>{{cite news |url=https://www.zdnet.com/article/google-alphago-ai-clean-sweeps-european-go-champion/ |title=Google AlphaGo AI clean sweeps European Go champion |work=[[ZDNet]] |date=28 January 2016 |access-date=28 January 2016 |archive-date=29 January 2016 |archive-url=https://web.archive.org/web/20160129053215/http://www.zdnet.com/article/google-alphago-ai-clean-sweeps-european-go-champion/ |url-status=live }}</ref> In a similar matchup, AlphaGo running on multiple computers won all 500 games played against other Go programs, and 77% of games played against AlphaGo running on a single computer. The distributed version in October 2015 was using 1,202 CPUs and 176 GPUs.<ref name="DeepMindnature2016" />

===Match against Fan Hui===
In October 2015, the distributed version of AlphaGo defeated the European Go champion Fan Hui,<ref name=MetzWired2016>{{cite magazine|title = In Major AI Breakthrough, Google System Secretly Beats Top Player at the Ancient Game of Go|url = https://www.wired.com/2016/01/in-a-huge-breakthrough-googles-ai-beats-a-top-player-at-the-game-of-go/|magazine = WIRED|access-date = 1 February 2016|language = en-US|date =27 January 2016|last = Metz|first = Cade}}</ref> a [[Software:Go ranks and ratings|2-dan]] (out of 9 dan possible) professional, five to zero.<ref name="bbcgo">{{cite news |url=https://www.bbc.com/news/technology-35420579 |title=Google achieves AI 'breakthrough' by beating Go champion |date=27 January 2016 |work=BBC News |access-date=20 July 2018 |archive-date=2 December 2021 |archive-url=https://web.archive.org/web/20211202120959/https://www.bbc.com/news/technology-35420579 |url-status=live }}</ref><ref>{{cite web|url = http://www.britgo.org/files/2016/deepmind/BGJ174-AlphaGo.pdf|title = Special Computer Go insert covering the AlphaGo v Fan Hui match|access-date = 1 February 2016|publisher = British Go Journal|year = 2017|archive-date = 2 February 2016|archive-url = https://web.archive.org/web/20160202065347/http://www.britgo.org/files/2016/deepmind/BGJ174-AlphaGo.pdf|url-status = live}}</ref> This was the first time a computer Go program had beaten a professional human player on a full-sized board without handicap.<ref name="lemondego">{{cite news |url=http://www.lemonde.fr/pixels/article/2016/01/27/premiere-defaite-d-un-professionnel-du-go-contre-une-intelligence-artificielle_4854886_4408996.html |title=Première défaite d'un professionnel du go contre une intelligence artificielle |date=27 January 2016 |work=Le Monde |language=fr |access-date=28 January 2016 |archive-date=29 January 2016 |archive-url=https://web.archive.org/web/20160129103942/http://www.lemonde.fr/pixels/article/2016/01/27/premiere-defaite-d-un-professionnel-du-go-contre-une-intelligence-artificielle_4854886_4408996.html |url-status=live }}</ref> The announcement of the news was delayed until 27 January 2016 to coincide with the publication of a paper in the journal ''Nature''<ref name="DeepMindnature2016">{{Cite journal|title = Mastering the game of Go with deep neural networks and tree search|journal = Nature| issn= 0028-0836|pages = 484–489|volume = 529|issue = 7587|doi = 10.1038/nature16961|pmid = 26819042|first1 = David|last1 = Silver|first2 = Aja|last2 = Huang|author-link2=Aja Huang|first3 = Chris J.|last3 = Maddison|first4 = Arthur|last4 = Guez|first5 = Laurent|last5 = Sifre|first6 = George van den|last6 = Driessche|first7 = Julian|last7 = Schrittwieser|first8 = Ioannis|last8 = Antonoglou|first9 = Veda|last9 = Panneershelvam|first10= Marc|last10= Lanctot|first11= Sander|last11= Dieleman|first12=Dominik|last12= Grewe|first13= John|last13= Nham|first14= Nal|last14= Kalchbrenner|first15= Ilya|last15= Sutskever|author-link15=Ilya Sutskever|first16= Timothy|last16= Lillicrap|first17= Madeleine|last17= Leach|first18= Koray|last18= Kavukcuoglu|first19= Thore|last19= Graepel|first20= Demis |last20=Hassabis|author-link20=Demis Hassabis|date= 28 January 2016|bibcode = 2016Natur.529..484S|s2cid = 515925}}{{closed access}}</ref> describing the algorithms used.<ref name="bbcgo" />

===Match against Lee Sedol===
AlphaGo played South Korean professional Go player Lee Sedol, ranked 9-dan, one of the best players at Go,<ref name="CNN0128" />{{Update after|2016|04|01}} with five games taking place at the Four Seasons Hotel in [[Place:Seoul|Seoul]], South Korea on 9, 10, 12, 13, and 15 March 2016,<ref>{{cite news|url=https://www.theguardian.com/technology/2016/feb/05/google-ai-alphago-world-no-1-lee-se-dol-live-broadcast|title=Google's AI AlphaGo to take on world No 1 Lee Sedol in live broadcast|newspaper={{wipe|The Guardian}}|date=5 February 2016|access-date=15 February 2016|archive-date=14 August 2017|archive-url=https://web.archive.org/web/20170814023821/https://www.theguardian.com/technology/2016/feb/05/google-ai-alphago-world-no-1-lee-se-dol-live-broadcast|url-status=live}}</ref><ref>{{cite web|url=http://www.businessinsider.com/google-deepmind-to-play-go-against-lee-sedol-in-south-korea-four-seasons-2016-2?r=UK&IR=T|title=Google DeepMind is going to take on the world's best Go player in a luxury 5-star hotel in South Korea|website=Business Insider|date=22 February 2016|access-date=23 February 2016|archive-date=2 March 2016|archive-url=https://web.archive.org/web/20160302114405/http://www.businessinsider.com/google-deepmind-to-play-go-against-lee-sedol-in-south-korea-four-seasons-2016-2?r=UK&IR=T|url-status=live}}</ref> which were video-streamed live.<ref>{{cite web|title = YouTube will livestream Google's AI playing Go superstar Lee Sedol in March|url = https://venturebeat.com/2016/02/04/youtube-will-livestream-googles-ai-playing-go-superstar-lee-sedol-in-march/|website = VentureBeat|access-date = 7 February 2016|date = 4 February 2016|last = Novet|first = Jordan|archive-date = 9 February 2016|archive-url = https://web.archive.org/web/20160209013817/http://venturebeat.com/2016/02/04/youtube-will-livestream-googles-ai-playing-go-superstar-lee-sedol-in-march/|url-status = live}}</ref>  Out of five games, AlphaGo won four games and Lee won the fourth game which made him recorded as the only human player who beat AlphaGo in all of its 74 official games.<ref name="koreatimes beatable">{{cite news|author1=Yoon Sung-won|title=Lee Se-dol shows AlphaGo beatable|url=https://www.koreatimes.co.kr/www/news/tech/2016/03/133_200267.html|access-date=15 March 2016|work=The Korea Times|date=14 March 2016|archive-date=14 March 2016|archive-url=https://web.archive.org/web/20160314184000/http://www.koreatimes.co.kr/www/news/tech/2016/03/133_200267.html|url-status=live}}</ref> AlphaGo ran on Google's cloud computing with its servers located in the United States.<ref name="JoongAng Ilbo">{{cite news|url=http://chinese.joins.com/gb/article.do?method=detail&art_id=148225&category=005001|title=李世乭：即使Alpha Go得到升级也一样能赢|newspaper=JoongAng Ilbo|date=23 February 2016|access-date=24 February 2016|language=zh|archive-date=4 March 2016|archive-url=https://web.archive.org/web/20160304010903/http://chinese.joins.com/gb/article.do?method=detail&art_id=148225&category=005001|url-status=live}}</ref> The match used Chinese rules with a 7.5-point [[Software:Komidashi|komi]], and each side had two hours of thinking time plus three 60-second byoyomi periods.<ref name="Korea Baduk Association"/> The version of AlphaGo playing against Lee used a similar amount of computing power as was used in the Fan Hui match.<ref>{{cite tweet |user=demishassabis|number=708488229750591488|date=11 March 2016|title=We are using roughly same amount of compute power as in Fan Hui match: distributing search over further machines has diminishing returns|author=Demis Hassabis|access-date=14 March 2016}}</ref> ''[[Finance:The Economist|The Economist]]'' reported that it used 1,920 [[Central processing unit|CPUs]] and 280 [[Graphics processing unit|GPUs]].<ref>{{cite news|url=https://www.economist.com/news/science-and-technology/21694540-win-or-lose-best-five-battle-contest-another-milestone|title=Showdown|newspaper=The Economist|access-date=19 November 2016|archive-date=14 August 2017|archive-url=https://web.archive.org/web/20170814060437/https://www.economist.com/news/science-and-technology/21694540-win-or-lose-best-five-battle-contest-another-milestone|url-status=live}}</ref> At the time of play, Lee Sedol had the second-highest number of Go international championship victories in the world after South Korean player Lee Changho who kept the world championship title for 16 years.<ref>{{cite news|author1=Steven Borowiec|title=Google's AI machine v world champion of 'Go': everything you need to know|url=https://www.theguardian.com/technology/2016/mar/09/googles-ai-machine-v-world-champion-of-go-everything-you-need-to-know|access-date=15 March 2016|work=The Guardian|date=9 March 2016|archive-date=15 March 2016|archive-url=https://web.archive.org/web/20160315002725/http://www.theguardian.com/technology/2016/mar/09/googles-ai-machine-v-world-champion-of-go-everything-you-need-to-know|url-status=live}}</ref> Since there is no single official method of [[Software:Go ranks and ratings#Rating systems|ranking in international Go]], the rankings may vary among the sources. While he was ranked top sometimes, some sources ranked Lee Sedol as the fourth-best player in the world at the time.<ref name="goratings.org 2016">{{cite web|author1=Rémi Coulom|title=Rating List of 2016-01-01|url=http://www.goratings.org/history/2016-01-01.html|access-date=18 March 2016|archive-url=https://web.archive.org/web/20160318041922/http://www.goratings.org/history/2016-01-01.html|archive-date=18 March 2016|author1-link=Rémi Coulom}}</ref><ref>{{cite news|title=Korean Go master proves human intuition still powerful in Go|url=http://digital.asiaone.com/digital/news/korean-go-master-proves-human-intuition-still-powerful-go|access-date=15 March 2016|work=The Korean Herald/ANN|date=14 March 2016|archive-url=https://web.archive.org/web/20160412234130/http://digital.asiaone.com/digital/news/korean-go-master-proves-human-intuition-still-powerful-go|archive-date=12 April 2016|url-status=dead}}</ref> AlphaGo was not specifically trained to face Lee nor was designed to compete with any specific human players.

The first three games were won by AlphaGo following resignations by Lee.<ref>{{cite news|url=https://www.bbc.co.uk/news/technology-35761246|title=Google's AI beats world Go champion in first of five matches – BBC News|work=BBC Online|access-date=9 March 2016|archive-date=10 March 2018|archive-url=https://web.archive.org/web/20180310103900/http://www.bbc.co.uk/news/technology-35761246|url-status=live}}</ref><ref>{{cite news|url=https://www.bbc.co.uk/news/technology-35771705|title=Google AI wins second Go game against world champion – BBC News|work=BBC Online|access-date=10 March 2016|archive-date=10 March 2016|archive-url=https://web.archive.org/web/20160310102208/http://www.bbc.co.uk/news/technology-35771705|url-status=live}}</ref> However, Lee beat AlphaGo in the fourth game, winning by resignation at move 180. AlphaGo then continued to achieve a fourth win, winning the fifth game by resignation.<ref>{{cite web|url=https://www.engadget.com/2016/03/14/the-final-lee-sedol-vs-alphago-match-is-about-to-start/|title=Google DeepMind AI wins final Go match for 4–1 series win|website=Engadget|access-date=15 March 2016|archive-date=15 March 2016|archive-url=https://web.archive.org/web/20160315044923/http://www.engadget.com/2016/03/14/the-final-lee-sedol-vs-alphago-match-is-about-to-start/|url-status=live}}</ref>

The prize was US$1 million. Since AlphaGo won four out of five and thus the series, the prize will be donated to charities, including UNICEF.<ref>{{cite news|url=https://www.boston25news.com/news/human-champion-certain-hell-beat-ai-at-ancient-chinese-game/105803732|title=Human champion certain he'll beat AI at ancient Chinese game|agency=Associated Press|date=22 February 2016|access-date=22 February 2016|archive-date=24 January 2019|archive-url=https://web.archive.org/web/20190124102048/https://www.boston25news.com/news/human-champion-certain-hell-beat-ai-at-ancient-chinese-game/105803732|url-status=dead}}</ref> Lee Sedol received $150,000 for participating in all five games and an additional $20,000 for his win in Game 4.<ref name="Korea Baduk Association">{{cite web|url=http://www.baduk.or.kr/news/report_view.asp?news_no=1671|title=이세돌 vs 알파고, '구글 딥마인드 챌린지 매치' 기자회견 열려|publisher=Korea Baduk Association|date=22 February 2016|access-date=22 February 2016|language=ko|archive-url=https://web.archive.org/web/20160303210212/http://www.baduk.or.kr/news/report_view.asp?news_no=1671|archive-date=3 March 2016|url-status=dead|df=dmy-all}}</ref>

In June 2016, at a presentation held at a university in the Netherlands, Aja Huang, one of the Deep Mind team, revealed that they had patched the logical weakness that occurred during the 4th game of the match between AlphaGo and Lee, and that after move 78 (which was dubbed the "divine move" by many professionals), it would play as intended and maintain Black's advantage. Before move 78, AlphaGo was leading throughout the game, but Lee's move caused the program's computing powers to be diverted and confused.<ref>{{Cite magazine|url=https://www.wired.com/2016/03/two-moves-alphago-lee-sedol-redefined-future/|title=In Two Moves, AlphaGo and Lee Sedol Redefined the Future|magazine=WIRED|access-date=2017-11-12|language=en-US}}</ref> Huang explained that AlphaGo's policy network of finding the most accurate move order and continuation did not precisely guide AlphaGo to make the correct continuation after move 78, since its value network did not determine Lee's 78th move as being the most likely, and therefore when the move was made AlphaGo could not make the right adjustment to the logical continuation.<ref name="Aja Huang Conference">{{cite web|url=http://weiqi.sports.tom.com/2016-07-08/00UP/13210553.html|title=黄士杰：AlphaGo李世石人机大战第四局问题已解决date=8 July 2016|access-date=8 July 2016|language=zh|archive-date=10 October 2018|archive-url=https://web.archive.org/web/20181010181235/http://weiqi.sports.tom.com/2016-07-08/00UP/13210553.html|url-status=dead}}</ref>

===Sixty online games===
{{Main|Software:Master}}
On 29 December 2016, a new account on the Tygem server named "Magister" (shown as 'Magist' at the server's Chinese version) from South Korea began to play games with professional players. It changed its account name to "Master" on 30 December, then moved to the FoxGo server on 1 January 2017. On 4 January, DeepMind confirmed that the "Magister" and the "Master" were both played by an updated version of AlphaGo, called AlphaGo Master.<ref name="Master">{{cite web|author=Demis Hassabis|url=https://twitter.com/demishassabis/status/816660463282954240|title=Demis Hassabis on Twitter: "Excited to share an update on #AlphaGo!"|publisher=Demis Hassabis's [[Company:Twitter|Twitter]] account|date=4 January 2017|access-date=4 January 2017|archive-date=4 May 2019|archive-url=https://web.archive.org/web/20190504005638/https://twitter.com/demishassabis/status/816660463282954240|url-status=live}}</ref><ref name="Nature-Master">{{cite journal|author=Elizabeth Gibney|title=Google reveals secret test of AI bot to beat top Go players|journal= Nature|date=4 January 2017|volume=541|issue=7636|page=142|doi=10.1038/nature.2017.21253|pmid=28079098|bibcode=2017Natur.541..142G|doi-access=free}}</ref> As of 5 January 2017, AlphaGo Master's online record was 60 wins and 0 losses,<ref name="WSJ Mourn">{{cite news|title=Humans Mourn Loss After Google Is Unmasked as China's Go Master|url=https://www.wsj.com/articles/ai-program-vanquishes-human-players-of-go-in-china-1483601561|access-date=6 January 2017|work=Wall Street Journal|date=5 January 2017|archive-date=26 May 2019|archive-url=https://web.archive.org/web/20190526022838/https://www.wsj.com/articles/ai-program-vanquishes-human-players-of-go-in-china-1483601561|url-status=live}}</ref> including three victories over Go's top-ranked player, Ke Jie,<ref>{{cite news|title=The world's best Go player says he still has "one last move" to defeat Google's AlphaGo AI|url=https://qz.com/878503/ke-jie-the-worlds-best-go-player-says-he-still-has-one-last-move-to-defeat-google-goog-deepminds-alphago-ai/|access-date=6 January 2017|work=Quartz|date=4 January 2017|archive-date=19 November 2020|archive-url=https://web.archive.org/web/20201119012110/https://qz.com/878503/ke-jie-the-worlds-best-go-player-says-he-still-has-one-last-move-to-defeat-google-goog-deepminds-alphago-ai/|url-status=live}}</ref> who had been quietly briefed in advance that Master was a version of AlphaGo.<ref name="WSJ Mourn" /> After losing to Master, Gu Li offered a bounty of 100,000 [[Finance:Renminbi|yuan]] (US$14,400) to the first human player who could defeat Master.<ref name="Nature-Master" /> Master played at the pace of 10 games per day. Many quickly suspected it to be an AI player due to little or no resting between games. Its adversaries included many world champions such as Ke Jie, Park Jeong-hwan, Yuta Iyama, Tuo Jiaxi, Mi Yuting, Shi Yue, Chen Yaoye, Li Qincheng, Gu Li, Chang Hao, Tang Weixing, Fan Tingyu, Zhou Ruiyang, Jiang Weijie, Chou Chun-hsun, Kim Ji-seok, Kang Dong-yun, Park Yeong-hun, and Won Seong-jin; national champions or world championship runners-up such as Lian Xiao, Tan Xiao, Meng Tailing, Dang Yifei, Huang Yunsong, Yang Dingxin, Gu Zihao, Shin Jinseo, Cho Han-seung, and An Sungjoon. All 60 games except one were fast-paced games with three 20 or 30 seconds byo-yomi. Master offered to extend the byo-yomi to one minute when playing with Nie Weiping in consideration of his age. After winning its 59th game Master revealed itself in the chatroom to be controlled by Dr. Aja Huang of the DeepMind team,<ref>{{cite web|url=http://www.thepaper.cn/newsDetail_forward_1593503|title=横扫中日韩棋手斩获59胜的Master发话：我是阿尔法狗|publisher=澎湃新闻|date=4 January 2017|access-date=11 December 2017|language=zh|archive-date=30 September 2020|archive-url=https://web.archive.org/web/20200930214237/http://www.thepaper.cn/newsDetail_forward_1593503|url-status=live}}</ref> then changed its nationality to the United Kingdom. After these games were completed, the co-founder of [[Company:DeepMind|DeepMind]], [[Biography:Demis Hassabis|Demis Hassabis]], said in a tweet, "we're looking forward to playing some official, full-length games later [2017] in collaboration with Go organizations and experts".<ref name="Master" /><ref name="Nature-Master" />

Go experts were impressed by the program's performance and its nonhuman play style; Ke Jie stated that "After humanity spent thousands of years improving our tactics, computers tell us that humans are completely wrong... I would go as far as to say not a single human has touched the edge of the truth of Go."<ref name="WSJ Mourn" />

===Future of Go Summit===
{{main|Software:Future of Go Summit}}
In the Future of Go Summit held in Wuzhen in May 2017, AlphaGo Master played three games with Ke Jie, the world No.1 ranked player, as well as two games with several top Chinese professionals, one pair Go game and one against a collaborating team of five human players.<ref>{{cite web|url=https://deepmind.com/blog/exploring-mysteries-alphago/|title=Exploring the mysteries of Go with AlphaGo and China's top players|date=2017-04-10|access-date=10 April 2017|archive-date=11 April 2017|archive-url=https://web.archive.org/web/20170411054218/https://deepmind.com/blog/exploring-mysteries-alphago/|url-status=dead}}</ref>

Google DeepMind offered 1.5 million dollar winner prizes for the three-game match between Ke Jie and Master while the losing side took 300,000 dollars.<ref>{{cite web|url=http://news.xinhuanet.com/english/2017-04/10/c_136197270.htm|title=World No.1 Go player Ke Jie takes on upgraded AlphaGo in May|date=2017-04-10|access-date=27 May 2017|archive-date=15 April 2017|archive-url=https://web.archive.org/web/20170415161356/http://news.xinhuanet.com/english/2017-04/10/c_136197270.htm|url-status=live}}</ref><ref>{{cite web|url=http://chuansong.me/n/1840585451964|title=Ke Jie vs. AlphaGo: 8 things you must know|date=2017-05-27|access-date=27 May 2017|archive-date=14 December 2017|archive-url=https://web.archive.org/web/20171214015725/http://chuansong.me/n/1840585451964|url-status=live}}</ref> Master won all three games against Ke Jie,<ref>{{cite magazine|url=https://www.wired.com/2017/05/revamped-alphago-wins-first-game-chinese-go-grandmaster/|title=Revamped AlphaGo Wins First Game Against Chinese Go Grandmaster|magazine=Wired|date=2017-05-23|last1=Metz|first1=Cade}}</ref><ref>{{cite magazine|url=https://www.wired.com/2017/05/googles-alphago-continues-dominance-second-win-china/|title=Google's AlphaGo Continues Dominance With Second Win in China|magazine=Wired|date=2017-05-25|last1=Metz|first1=Cade}}</ref> after which AlphaGo was awarded professional 9-dan by the Chinese Weiqi Association.<ref name="AlphaGo 9-dan"/>

After winning its three-game match against Ke Jie, the top-rated world Go player, AlphaGo retired. DeepMind also disbanded the team that worked on the game to focus on AI research in other areas.<ref name="AlphaGo's Designers Explore New AI"/> After the Summit, Deepmind published 50 full length AlphaGo vs AlphaGo matches, as a gift to the Go community.<ref name="50 games">{{cite web|url=https://deepmind.com/research/alphago/alphago-vs-alphago-self-play-games/|title=Full length games for Go players to enjoy|publisher=Deepmind|access-date=2017-05-28|archive-date=5 August 2019|archive-url=https://web.archive.org/web/20190805063354/https://deepmind.com/research/alphago/alphago-vs-alphago-self-play-games/|url-status=dead}}</ref>

===AlphaGo Zero and AlphaZero===
{{Main|AlphaGo Zero|AlphaZero}}
AlphaGo's team published an article in the journal ''Nature'' on 19 October 2017, introducing AlphaGo Zero, a version without human data and stronger than any previous human-champion-defeating version.<ref name="Nature2017">{{cite journal|first1= David|last1= Silver|first2= Julian|last2= Schrittwieser|first3= Karen|last3= Simonyan|first4= Ioannis|last4= Antonoglou|first5= Aja|last5= Huang|author-link5= Aja Huang|first6= Arthur|last6= Guez|first7= Thomas|last7= Hubert|first8= Lucas|last8= Baker|first9= Matthew|last9= Lai|first10= Adrian|last10= Bolton|first11= Yutian|last11= Chen|author-link11= Chen Yutian|first12= Timothy|last12= Lillicrap|first13= Hui|last13= Fan|author-link13= Fan Hui|first14= Laurent|last14= Sifre|first15= George van den|last15= Driessche|first16= Thore|last16= Graepel|first17= Demis|last17= Hassabis|author-link17= Demis Hassabis|title= Mastering the game of Go without human knowledge|journal= Nature|issn= 0028-0836|pages= 354–359|volume= 550|issue= 7676|doi= 10.1038/nature24270|date= 19 October 2017|bibcode= 2017Natur.550..354S|pmid= 29052630|s2cid= 205261034|url= https://discovery.ucl.ac.uk/id/eprint/10045895/1/agz_unformatted_nature.pdf|access-date= 29 August 2020|archive-date= 24 November 2020|archive-url= https://web.archive.org/web/20201124151015/https://discovery.ucl.ac.uk/id/eprint/10045895/1/agz_unformatted_nature.pdf|url-status= live}}{{closed access}}</ref> By playing games against itself, AlphaGo Zero surpassed the strength of AlphaGo Lee in three days by winning 100 games to 0, reached the level of AlphaGo Master in 21 days, and exceeded all the old versions in 40 days.<ref name="Deepmind20171018">{{cite web|url=https://deepmind.com/blog/alphago-zero-learning-scratch/|title=AlphaGo Zero: Learning from scratch|publisher=[[Company:DeepMind|DeepMind]] official website|date=18 October 2017|access-date=19 October 2017|archive-date=19 October 2017|archive-url=https://web.archive.org/web/20171019220157/https://deepmind.com/blog/alphago-zero-learning-scratch/|url-status=dead}}</ref>

In a paper released on [[ArXiv|arXiv]] on 5 December 2017, DeepMind claimed that it generalized AlphaGo Zero's approach into a single AlphaZero algorithm, which achieved within 24 hours a superhuman level of play in the games of chess, shogi, and [[Software:Go (game)|Go]] by defeating world-champion programs, [[Software:Stockfish (chess)|Stockfish]], [[Software:Elmo (shogi engine)|Elmo]], and 3-day version of AlphaGo Zero in each case.<ref>{{Cite arXiv|first1=David|last1= Silver|first2=Thomas|last2= Hubert|first3= Julian|last3=Schrittwieser|first4= Ioannis|last4=Antonoglou |first5= Matthew|last5= Lai|first6= Arthur|last6= Guez|first7= Marc|last7= Lanctot|first8= Laurent|last8= Sifre|first9= Dharshan|last9= Kumaran|first10= Thore|last10= Graepel|first11= Timothy|last11= Lillicrap|first12= Karen|last12= Simonyan|first13=Demis |last13=Hassabis|author-link13=Demis Hassabis |eprint=1712.01815|title=Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm|class=cs.AI|date=5 December 2017}}</ref>

===Teaching tool===
On 11 December 2017, DeepMind released AlphaGo teaching tool on its website<ref>{{cite web|url=https://alphagoteach.deepmind.com/|title=AlphaGo teaching tool|publisher=[[Company:DeepMind|DeepMind]]|access-date=11 December 2017|archive-date=12 December 2017|archive-url=https://web.archive.org/web/20171212143750/https://alphagoteach.deepmind.com/|url-status=live}}</ref> to analyze winning rates of different Go openings as calculated by AlphaGo Master.<ref name="sina20171211">{{cite web|url=http://sports.sina.com.cn/go/2017-12-11/doc-ifypnsip8212788.shtml|title=AlphaGo教学工具上线 樊麾：使用Master版本|publisher=Sina.com.cn|date=11 December 2017|access-date=11 December 2017|language=zh|archive-date=12 December 2017|archive-url=https://web.archive.org/web/20171212032055/http://sports.sina.com.cn/go/2017-12-11/doc-ifypnsip8212788.shtml|url-status=live}}</ref> The teaching tool collects 6,000 Go openings from 230,000 human games each analyzed with 10,000,000 simulations by AlphaGo Master. Many of the openings include human move suggestions.<ref name="sina20171211"/>

==Versions==
An early version of AlphaGo was tested on hardware with various numbers of [[Central processing unit|CPU]]s and [[Graphics processing unit|GPU]]s, running in asynchronous or distributed mode. Two seconds of thinking time was given to each move. The resulting Elo ratings are listed below.<ref name="DeepMindnature2016"/> In the matches with more time per move higher ratings are achieved.

{| class="wikitable sortable" style="text-align:center"
|+ Configuration and performance
|- style="background:#ececec; vertical-align:top;" 
!Configuration !!Search<br>threads!!No. of CPU!!No. of GPU!!Elo rating
|-
| align=left|Single<ref name="DeepMindnature2016"/> <sup>p.&nbsp;10–11</sup> ||40||48||1||2,181
|-
| align=left|Single||40||48||2||2,738
|-
| align=left|Single||40||48||4||2,850
|-
| align=left|Single||40||48||8||2,890
|-
| align=left|[[Distributed computing|Distributed]]||12||428||64||2,937
|-
| align=left|Distributed||24||764||112||3,079
|-
| align=left|Distributed||40||1,202||176||3,140
|-
| align=left|Distributed||64||1,920||280||3,168
|}

In May 2016, Google unveiled its own proprietary hardware "[[Engineering:Tensor processing unit|tensor processing unit]]s", which it stated had already been deployed in multiple internal projects at Google, including the AlphaGo match against Lee Sedol.<ref>{{cite news|last1=McMillan|first1=Robert|title=Google Isn't Playing Games With New Chip|url=https://www.wsj.com/articles/google-isnt-playing-games-with-new-chip-1463597820|access-date=26 June 2016|work={{wipe|The Wall Street Journal}}|date=18 May 2016|archive-date=29 June 2016|archive-url=https://web.archive.org/web/20160629031433/http://www.wsj.com/articles/google-isnt-playing-games-with-new-chip-1463597820|url-status=live}}</ref><ref>{{cite web|url=https://cloudplatform.googleblog.com/2016/05/Google-supercharges-machine-learning-tasks-with-custom-chip.html|title=Google supercharges machine learning tasks with TPU custom chip|last=Jouppi|first=Norm|date=May 18, 2016|website=Google Cloud Platform Blog|language=en-US|access-date=2016-06-26|archive-date=18 May 2016|archive-url=https://web.archive.org/web/20160518201516/https://cloudplatform.googleblog.com/2016/05/Google-supercharges-machine-learning-tasks-with-custom-chip.html|url-status=live}}</ref>

In the [[Software:Future of Go Summit|Future of Go Summit]] in May 2017, DeepMind disclosed that the version of AlphaGo used in this Summit was [[Software:Master|AlphaGo Master]],<ref name="sina0524">{{cite web|url=http://sports.sina.com.cn/go/2017-05-24/doc-ifyfkqwe0969707.shtml?cre=tagspc&mod=g&r=user&pos=5_7|title=AlphaGo官方解读让三子 对人类高手没这种优势|publisher=Sina|date=25 May 2017|access-date=2 June 2017|language=zh|archive-date=16 April 2021|archive-url=https://web.archive.org/web/20210416081136/http://sports.sina.com.cn/go/2017-05-24/doc-ifyfkqwe0969707.shtml?cre=tagspc&mod=g&r=user&pos=5_7|url-status=live}}</ref><ref>{{cite web|url=http://sports.sina.com.cn/go/2017-05-24/doc-ifyfkqwe0899285.shtml|title=各版alphago实力对比 master能让李世石版3子|publisher=Sina|date=24 May 2017|access-date=2 June 2017|language=zh|archive-date=3 June 2017|archive-url=https://web.archive.org/web/20170603035444/http://sports.sina.com.cn/go/2017-05-24/doc-ifyfkqwe0899285.shtml|url-status=live}}</ref> and revealed that it had measured the strength of different versions of the software. AlphaGo Lee, the version used against Lee, could give AlphaGo Fan, the version used in AlphaGo vs. Fan Hui, three stones, and AlphaGo Master was even three stones stronger.<ref>{{cite web|url=http://www.usgo.org/news/2017/05/new-version-of-alphago-self-trained-and-much-more-efficient/|title=New version of AlphaGo self-trained and much more efficient|publisher=American Go Association|date=24 May 2017|access-date=1 June 2017|archive-date=3 June 2017|archive-url=https://web.archive.org/web/20170603192500/http://www.usgo.org/news/2017/05/new-version-of-alphago-self-trained-and-much-more-efficient/|url-status=live}}</ref>
{| class="wikitable sortable" style="text-align:center"
|+ Configuration and strength<ref name="sohu0524">{{cite web|url=http://www.sohu.com/a/143092581_473283|title=【柯洁战败解密】AlphaGo Master最新架构和算法，谷歌云与TPU拆解|publisher=Sohu|date=24 May 2017|access-date=1 June 2017|language=zh|archive-date=17 September 2017|archive-url=https://web.archive.org/web/20170917171454/https://www.sohu.com/a/143092581_473283|url-status=live}}</ref>
|- style="background:#ececec; vertical-align:top;" 
!Versions!!Hardware!!Elo rating!!Date!!Results
|-
| align=left|AlphaGo Fan||176 [[Graphics processing unit|GPUs]],<ref name="Deepmind20171018"/> distributed||3,144<ref name="Nature2017"/> ||Oct 2015||5:0 against Fan Hui
|-
| align=left|AlphaGo Lee|| 48 [[Engineering:Tensor processing unit|TPUs]],<ref name="Deepmind20171018"/> distributed||3,739<ref name="Nature2017"/>||Mar 2016||4:1 against Lee Sedol
|-
| align=left|[[Software:Master|AlphaGo Master]]||4 TPUs,<ref name="Deepmind20171018"/> single machine||4,858<ref name="Nature2017"/>||May 2017|| 60:0 against professional players;<br/>[[Software:Future of Go Summit|Future of Go Summit]]
|-
| align=left|[[AlphaGo Zero]] (40 block)||4 TPUs,<ref name="Deepmind20171018"/> single machine||5,185<ref name="Nature2017"/>||Oct 2017|| 100:0 against AlphaGo Lee
89:11 against AlphaGo Master
|-
| align=left|[[AlphaZero]] (20 block)||4 TPUs, single machine||5,018<ref>{{Cite journal|url=https://www.science.org/doi/full/10.1126/science.aar6404|title=A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play|first1=David|last1=Silver|first2=Thomas|last2=Hubert|first3=Julian|last3=Schrittwieser|first4=Ioannis|last4=Antonoglou|first5=Matthew|last5=Lai|first6=Arthur|last6=Guez|first7=Marc|last7=Lanctot|first8=Laurent|last8=Sifre|first9=Dharshan|last9=Kumaran|first10=Thore|last10=Graepel|first11=Timothy|last11=Lillicrap|first12=Karen|last12=Simonyan|first13=Demis|last13=Hassabis|date=7 December 2018|journal=Science|volume=362|issue=6419|pages=1140–1144|via=science.org (Atypon)|doi=10.1126/science.aar6404|pmid=30523106|bibcode=2018Sci...362.1140S|s2cid=54457125|access-date=13 October 2021|archive-date=9 October 2021|archive-url=https://web.archive.org/web/20211009112230/https://www.science.org/doi/full/10.1126/science.aar6404|url-status=live|doi-access=free}}</ref>

||Dec 2017||60:40 against AlphaGo Zero (20 block)
|}

==Algorithm==
As of 2016, AlphaGo's algorithm uses a combination of [[Machine learning|machine learning]] and tree search techniques, combined with extensive training, both from human and computer play. It uses [[Monte Carlo tree search]], guided by a "value network" and a "policy network", both implemented using deep neural network technology.<ref name="googlego" /><ref name="DeepMindnature2016" /> A limited amount of game-specific feature detection pre-processing (for example, to highlight whether a move matches a [[Software:List of Go terms#Nakade|nakade]] pattern) is applied to the input before it is sent to the neural networks.<ref name="DeepMindnature2016" /> The networks are convolutional neural networks with 12 layers, trained by [[Reinforcement learning|reinforcement learning]].<ref>{{cite journal|ref=none|last1=Silver|first1=David|last2=Schrittwieser|first2=Julian|last3=Simonyan|first3=Karen|last4=Antonoglou|first4=Ioannis|last5=Huang|first5=Aja|author-link5=Aja Huang|last6=Guez|first6=Arthur|last7=Hubert|first7=Thomas|last8=Baker|first8=Lucas|last9=Lai|first9=Matthew|last10=Bolton|first10=Adrian|last11=Chen|first11=Yutian|author-link11=Chen Yutian|last12=Lillicrap|first12=Timothy|last13=Fan|first13=Hui|author-link13=Fan Hui|last14=Sifre|first14=Laurent|last15=Driessche|first15=George van den|last16=Graepel|first16=Thore|last17=Hassabis|first17=Demis|author-link17=Demis Hassabis|date=19 October 2017|title=Mastering the game of Go without human knowledge|journal=Nature|volume=550|issue=7676|pages=354–359|url=http://discovery.ucl.ac.uk/10045895/1/agz_unformatted_nature.pdf|bibcode=2017Natur.550..354S|doi=10.1038/nature24270|issn=0028-0836|pmid=29052630|s2cid=205261034|quote=AlphaGo Lee... 12 convolutional layers|access-date=13 October 2021|archive-date=18 July 2018|archive-url=https://web.archive.org/web/20180718225914/http://discovery.ucl.ac.uk/10045895/1/agz_unformatted_nature.pdf|url-status=live}}</ref>

The system's neural networks were initially bootstrapped from human gameplay expertise. AlphaGo was initially trained to mimic human play by attempting to match the moves of expert players from recorded historical games, using a database of around 30 million moves.<ref name=MetzWired2016/> Once it had reached a certain degree of proficiency, it was trained further by being set to play large numbers of games against other instances of itself, using [[Reinforcement learning|reinforcement learning]] to improve its play.<ref name="googlego"/> To avoid "disrespectfully" wasting its opponent's time, the program is specifically programmed to resign if its assessment of win probability falls beneath a certain threshold; for the match against Lee, the resignation threshold was set to 20%.<ref>{{cite news|author1=Cade Metz|title=Go Grandmaster Lee Sedol Grabs Consolation Win Against Google's AI|url=https://www.wired.com/2016/03/go-grandmaster-lee-sedol-grabs-consolation-win-googles-ai/|access-date=29 March 2016|work=Wired News|date=13 March 2016|archive-date=17 November 2017|archive-url=https://web.archive.org/web/20171117001746/https://www.wired.com/2016/03/go-grandmaster-lee-sedol-grabs-consolation-win-googles-ai/|url-status=live}}</ref>

==Style of play==
Toby Manning, the match referee for AlphaGo vs. Fan Hui, has described the program's style as "conservative".<ref name=":0">{{cite journal|title = Google AI algorithm masters ancient game of Go|journal = Nature|date = 27 January 2016|last = Gibney|first = Elizabeth|volume = 529|issue = 7587|pages = 445–6|pmid = 26819021|doi = 10.1038/529445a|bibcode = 2016Natur.529..445G|doi-access = free}}</ref> AlphaGo's playing style strongly favours greater probability of winning by fewer points over lesser probability of winning by more points.<ref name="pcworld_unusual" /> Its strategy of maximising its probability of winning is distinct from what human players tend to do which is to maximise territorial gains, and explains some of its odd-looking moves.<ref>{{cite journal |url=http://www.nature.com/news/the-go-files-ai-computer-clinches-victory-against-go-champion-1.19553 |title=The Go Files: AI computer clinches victory against Go champion |first=Tanguy |last=Chouard |journal=Nature |date=12 March 2016 |doi=10.1038/nature.2016.19553 |s2cid=155164502 |access-date=18 December 2016 |archive-date=18 June 2016 |archive-url=https://web.archive.org/web/20160618014809/http://www.nature.com/news/the-go-files-ai-computer-clinches-victory-against-go-champion-1.19553 |url-status=live }}</ref> It makes a lot of opening moves that have never or seldom been made by humans. It likes to use [[Software:List of Go terms#Shoulder hit|shoulder hits]], especially if the opponent is over concentrated.<ref>{{cite web|url=http://sports.sina.com.cn/go/2017-01-11/doc-ifxzqnim3941818.shtml|title=韩国研究新版AlphaGo：穿越而来展示未来围棋|publisher=Sina.com|date=11 January 2017|access-date=24 April 2017|language=Chinese|archive-date=24 April 2017|archive-url=https://web.archive.org/web/20170424174309/http://sports.sina.com.cn/go/2017-01-11/doc-ifxzqnim3941818.shtml|url-status=live}}</ref>

==Responses to 2016 victory ==

===AI community===
AlphaGo's March 2016 victory was a major milestone in artificial intelligence research.<ref name="latimes milestone">{{cite news|author1=Steven Borowiec|author2=Tracey Lien|title=AlphaGo beats human Go champ in milestone for artificial intelligence|url=https://www.latimes.com/world/asia/la-fg-korea-alphago-20160312-story.html|access-date=13 March 2016|work={{wipe|Los Angeles Times}}|date=12 March 2016|archive-date=13 May 2018|archive-url=https://web.archive.org/web/20180513234132/http://www.latimes.com/world/asia/la-fg-korea-alphago-20160312-story.html|url-status=live}}</ref> Go had previously been regarded as a hard problem in machine learning that was expected to be out of reach for the technology of the time.<ref name="latimes milestone" /><ref>{{cite news |title=A computer has beaten a professional at the world's most complex board game |url=https://www.independent.co.uk/life-style/gadgets-and-tech/news/google-alphago-computer-beats-professional-at-worlds-most-complex-board-game-go-a6837506.html |archive-url=https://web.archive.org/web/20160128012935/http://www.independent.co.uk/life-style/gadgets-and-tech/news/google-alphago-computer-beats-professional-at-worlds-most-complex-board-game-go-a6837506.html |archive-date=2016-01-28 |url-access=limited |url-status=live |newspaper=The Independent |access-date=28 January 2016 |date=27 January 2016 |last=Connor |first=Steve}}</ref><ref>{{cite news |title=Google's AI beats human champion at Go |url=http://www.cbc.ca/news/technology/alphago-ai-1.3422347 |work=CBC News |access-date=28 January 2016 |date=27 January 2016 |archive-date=10 March 2016 |archive-url=https://web.archive.org/web/20160310181352/http://www.cbc.ca/news/technology/alphago-ai-1.3422347 |url-status=live }}</ref> Most experts thought a Go program as powerful as AlphaGo was at least five years away;<ref>{{cite news|author1=Dave Gershgorn|title=GOOGLE'S ALPHAGO BEATS WORLD CHAMPION IN THIRD MATCH TO WIN ENTIRE SERIES|url=http://www.popsci.com/googles-alphago-beats-world-champion-in-third-match-to-win-entire-series|access-date=13 March 2016|work=Popular Science|date=12 March 2016|archive-date=16 December 2016|archive-url=https://web.archive.org/web/20161216144211/http://www.popsci.com/googles-alphago-beats-world-champion-in-third-match-to-win-entire-series|url-status=live}}</ref> some experts thought that it would take at least another decade before computers would beat Go champions.<ref name="DeepMindnature2016" /><ref name="cbc sweeps">{{cite news|title=Google DeepMind computer AlphaGo sweeps human champ in Go matches|url=http://www.cbc.ca/news/technology/go-google-alphago-lee-sedol-deepmind-1.3488913|access-date=13 March 2016|work=CBC News|agency=Associated Press|date=12 March 2016|archive-date=13 March 2016|archive-url=https://web.archive.org/web/20160313002739/http://www.cbc.ca/news/technology/go-google-alphago-lee-sedol-deepmind-1.3488913|url-status=live}}</ref><ref>{{cite news|author1=Sofia Yan|title=A Google computer victorious over the world's 'Go' champion|url=https://money.cnn.com/2016/03/12/technology/google-deepmind-alphago-wins/|access-date=13 March 2016|work=CNN Money|date=12 March 2016|archive-date=8 August 2020|archive-url=https://web.archive.org/web/20200808093512/https://money.cnn.com/2016/03/12/technology/google-deepmind-alphago-wins/|url-status=live}}</ref> Most observers at the beginning of the 2016 matches expected Lee to beat AlphaGo.<ref name="latimes milestone" />

With games such as checkers (that has been "[[Solved game|solved]]" by the Chinook draughts player team), chess, and now Go won by computers, victories at popular board games can no longer serve as major milestones for artificial intelligence in the way that they used to.<!-- <ref name="latimes milestone" /> --> Deep Blue's [[Biography:Murray Campbell|Murray Campbell]] called AlphaGo's victory "the end of an era... board games are more or less done and it's time to move on."<ref name="latimes milestone" />

When compared with Deep Blue or [[Engineering:Watson (computer)|Watson]], AlphaGo's underlying algorithms are potentially more general-purpose and may be evidence that the scientific community is making progress towards [[Artificial general intelligence|artificial general intelligence]].<ref name="pcworld_unusual" /><ref>{{cite news|title=AlphaGo: Google's artificial intelligence to take on world champion of ancient Chinese board game|url=http://www.abc.net.au/news/2016-03-08/google-artificial-intelligence-to-face-board-game-champion/7231192|access-date=13 March 2016|work=Australian Broadcasting Corporation|date=8 March 2016|archive-date=15 June 2016|archive-url=https://web.archive.org/web/20160615055816/http://www.abc.net.au/news/2016-03-08/google-artificial-intelligence-to-face-board-game-champion/7231192|url-status=live}}</ref> Some commentators believe AlphaGo's victory makes for a good opportunity for society to start preparing for the possible future impact of [[Artificial general intelligence|machines with general purpose intelligence]]. As noted by entrepreneur Guy Suter, AlphaGo only knows how to play Go and doesn't possess general-purpose intelligence; "[It] couldn't just wake up one morning and decide it wants to learn how to use firearms."<ref name="latimes milestone" /> AI researcher [[Biography:Stuart J. Russell|Stuart Russell]] said that AI systems such as AlphaGo have progressed quicker and become more powerful than expected, and we must therefore develop methods to ensure they "remain under human control".<ref name="phys.org eye">{{cite news|author1=Mariëtte Le Roux|title=Rise of the Machines: Keep an eye on AI, experts warn|url=http://phys.org/news/2016-03-machines-eye-ai-experts.html|access-date=13 March 2016|work=Phys.org|date=12 March 2016|archive-date=13 March 2016|archive-url=https://web.archive.org/web/20160313104358/http://phys.org/news/2016-03-machines-eye-ai-experts.html|url-status=live}}</ref> Some scholars, <!-- delete Hawking part? as not "response" to victories, nor is he part of "AI community" (or Go)--> such as [[Biography:Stephen Hawking|Stephen Hawking]], warned (in May 2015 before the matches) that some future self-improving AI could gain actual general intelligence, leading to an unexpected [[Unsolved:AI takeover|AI takeover]];<!-- <ref name="phys.org challenge" /> --> other scholars disagree: AI expert Jean-Gabriel Ganascia believes that "Things like '[[Philosophy:Common sense|common sense]]'... may never be reproducible",<ref name="phys.org challenge">{{cite news|author1=Mariëtte Le Roux|author2=Pascale Mollard|title=Game over? New AI challenge to human smarts (Update)|url=http://phys.org/news/2016-03-game-ai-human-smarts.html|access-date=13 March 2016|work=phys.org|date=8 March 2016|archive-date=14 March 2016|archive-url=https://web.archive.org/web/20160314032122/http://phys.org/news/2016-03-game-ai-human-smarts.html|url-status=live}}</ref> and says "I don't see why we would speak about fears. On the contrary, this raises hopes in many domains such as health and space exploration."<ref name="phys.org eye" /> Computer scientist [[Biography:Richard S. Sutton|Richard Sutton]] said "I don't think people should be scared... but I do think people should be paying attention."<ref>{{cite news|author1=Tanya Lewis|title=An AI expert says Google's Go-playing program is missing 1 key feature of human intelligence|url=http://www.businessinsider.com/what-does-googles-deepmind-victory-mean-for-ai-2016-3|access-date=13 March 2016|work=Business Insider|date=11 March 2016|archive-date=12 March 2016|archive-url=https://web.archive.org/web/20160312160057/http://www.businessinsider.com/what-does-googles-deepmind-victory-mean-for-ai-2016-3|url-status=live}}</ref>

In China, AlphaGo was a "[[Engineering:Sputnik crisis|Sputnik moment]]" which helped convince the Chinese government to prioritize and dramatically increase funding for artificial intelligence.<ref>{{cite news|last1=Mozur|first1=Paul|title=Beijing Wants A.I. to Be Made in China by 2030|url=https://www.nytimes.com/2017/07/20/business/china-artificial-intelligence.html|access-date=11 April 2018|work=The New York Times|date=20 July 2017|archive-date=11 April 2018|archive-url=https://web.archive.org/web/20180411174247/https://www.nytimes.com/2017/07/20/business/china-artificial-intelligence.html|url-status=live}}</ref>

In 2017, the DeepMind AlphaGo team received the inaugural IJCAI [[Biography:Marvin Minsky|Marvin Minsky]] medal for Outstanding Achievements in AI. "AlphaGo is a wonderful achievement, and a perfect example of what the Minsky Medal was initiated to recognise", said Professor [[Biography:Michael Wooldridge (computer scientist)|Michael Wooldridge]], Chair of the IJCAI Awards Committee.  "What particularly impressed IJCAI was that AlphaGo achieves what it does through a brilliant combination of classic AI techniques as well as the state-of-the-art machine learning techniques that DeepMind is so closely associated with. It's a breathtaking demonstration of contemporary AI, and we are delighted to be able to recognise it with this award."<ref>{{cite news|title=Marvin Minsky Medal for Outstanding Achievements in AI|url=http://www.ijcai.org/awards/minsky_medal|access-date=21 October 2017|work=International Joint Conference on Artificial Intelligence|date=19 October 2017|language=en|archive-date=21 October 2017|archive-url=https://web.archive.org/web/20171021080752/http://www.ijcai.org/awards/minsky_medal|url-status=live}}</ref>

===Go community===
Go is a popular game in China, Japan and Korea, and the 2016 matches were watched by perhaps a hundred million people worldwide.<ref name="latimes milestone" /><ref>{{cite news|author1=CHOE SANG-HUN|title=Google's Computer Program Beats Lee Se-dol in Go Tournament|url=https://www.nytimes.com/2016/03/16/world/asia/korea-alphago-vs-lee-sedol-go.html|access-date=18 March 2016|work={{wipe|The New York Times}}|date=16 March 2016|quote=More than 100 million people watched the AlphaGo-Lee matches, Mr. Hassabis said.|archive-date=18 March 2016|archive-url=https://web.archive.org/web/20160318070249/http://www.nytimes.com/2016/03/16/world/asia/korea-alphago-vs-lee-sedol-go.html|url-status=live}}</ref> Many top Go players characterized AlphaGo's unorthodox plays as seemingly-questionable moves that initially befuddled onlookers, but made sense in hindsight:<ref name="cbc sweeps" /> "All but the very best Go players craft their style by imitating top players. AlphaGo seems to have totally original moves it creates itself."<ref name="latimes milestone" /> AlphaGo appeared to have unexpectedly become much stronger, even when compared with its October 2015 match<ref>{{cite news|author1=John Ribeiro|title=Google's AlphaGo AI program strong but not perfect, says defeated South Korean Go player|url=http://www.pcworld.com/article/3043211/big-win-for-ai-as-google-alphago-program-trounces-korean-player-in-go-tournament.html|access-date=13 March 2016|work=PC World|date=12 March 2016|archive-date=13 March 2016|archive-url=https://web.archive.org/web/20160313095830/http://www.pcworld.com/article/3043211/big-win-for-ai-as-google-alphago-program-trounces-korean-player-in-go-tournament.html|url-status=live}}</ref> where a computer had beaten a Go professional for the first time ever without the advantage of a handicap.<ref name="nature react" /> The day after Lee's first defeat, Jeong Ahram, the lead Go correspondent for one of South Korea's biggest daily newspapers, said "Last night was very gloomy... Many people drank alcohol."<ref>{{cite news|title=How victory for Google's Go AI is stoking fear in South Korea|url=https://www.newscientist.com/article/2080927-how-victory-for-googles-go-ai-is-stoking-fear-in-south-korea/|access-date=18 March 2016|work=New Scientist|date=15 March 2016|first=Mark|last=Zastrow|archive-date=21 March 2016|archive-url=https://web.archive.org/web/20160321053851/https://www.newscientist.com/article/2080927-how-victory-for-googles-go-ai-is-stoking-fear-in-south-korea/|url-status=live}}</ref> The Korea Baduk Association, the organization that oversees Go professionals in South Korea, awarded AlphaGo an honorary 9-dan title for exhibiting creative skills and pushing forward the game's progress.<ref>{{cite news|author1=JEE HEUN KAHNG|author2=SE YOUNG LEE|title=Google artificial intelligence program beats S. Korean Go pro with 4–1 score|url=https://www.reuters.com/article/us-science-intelligence-go-idUSKCN0WH0XJ|access-date=18 March 2016|work={{wipe|Reuters}}|date=15 March 2016|archive-date=28 July 2017|archive-url=https://web.archive.org/web/20170728183907/http://www.reuters.com/article/us-science-intelligence-go-idUSKCN0WH0XJ|url-status=live}}</ref>

China's Ke Jie, an 18-year-old generally recognized as the world's best Go player at the time,<ref name="goratings.org 2016" /><ref name="telegraph china">{{cite news|author1=Neil Connor|title=Google AlphaGo 'can't beat me' says China Go grandmaster|url=https://www.telegraph.co.uk/news/worldnews/asia/china/12190917/Google-AlphaGo-cant-beat-me-says-China-Go-grandmaster.html|access-date=13 March 2016|work=The Telegraph (UK)|date=11 March 2016|archive-date=13 March 2016|archive-url=https://web.archive.org/web/20160313094230/http://www.telegraph.co.uk/news/worldnews/asia/china/12190917/Google-AlphaGo-cant-beat-me-says-China-Go-grandmaster.html|url-status=live}}</ref> initially claimed that he would be able to beat AlphaGo, but declined to play against it for fear that it would "copy my style".<ref name="telegraph china" /> As the matches progressed, Ke Jie went back and forth, stating that "it is highly likely that I (could) lose" after analysing the first three matches,<ref>{{cite web|url=http://english.donga.com/List/3/all/26/527586/1|title=Chinese Go master Ke Jie says he could lose to AlphaGo : The DONG-A ILBO|access-date=17 March 2016|archive-date=15 March 2016|archive-url=https://web.archive.org/web/20160315044611/http://english.donga.com/List/3/all/26/527586/1|url-status=live}}</ref> but regaining confidence after AlphaGo displayed flaws in the fourth match.<ref>{{cite web |url=http://m.hankooki.com/m_sp_view.php?WM=sp&FILE_NO=c3AyMDE2MDMxNDE4MDIzMDEzNjU3MC5odG0=&ref=search.naver.com |title=...if today's performance was its true capability, then it doesn't deserve to play against me. |publisher=M.hankooki.com |date=2016-03-14 |access-date=2018-06-05 |archive-date=15 March 2016 |archive-url=https://web.archive.org/web/20160315153117/http://m.hankooki.com/m_sp_view.php?WM=sp&FILE_NO=c3AyMDE2MDMxNDE4MDIzMDEzNjU3MC5odG0=&ref=search.naver.com |url-status=dead }}</ref>

Toby Manning, the referee of AlphaGo's match against Fan Hui, and Hajin Lee, secretary general of the International Go Federation, both reason that in the future, Go players will get help from computers to learn what they have done wrong in games and improve their skills.<ref name="nature react">{{cite journal |url=http://www.nature.com/news/go-players-react-to-computer-defeat-1.19255 |title=Go players react to computer defeat |first=Elizabeth |last=Gibney |journal=Nature |year=2016 |doi=10.1038/nature.2016.19255 |s2cid=146868978 |access-date=29 January 2016 |archive-date=30 January 2016 |archive-url=https://web.archive.org/web/20160130123457/http://www.nature.com/news/go-players-react-to-computer-defeat-1.19255 |url-status=live }}</ref>

After game two, Lee said he felt "speechless": "From the very beginning of the match, I could never manage an upper hand for one single move. It was AlphaGo's total victory."<ref>{{cite news|author1=CHOE SANG-HUN|title=In Seoul, Go Games Spark Interest (and Concern) About Artificial Intelligence|url=https://www.nytimes.com/2016/03/15/insider/in-seoul-authentic-enthusiasm-and-concern-over-artificial-intelligence.html|access-date=18 March 2016|work=The New York Times|date=15 March 2016|archive-date=18 March 2016|archive-url=https://web.archive.org/web/20160318124521/http://www.nytimes.com/2016/03/15/insider/in-seoul-authentic-enthusiasm-and-concern-over-artificial-intelligence.html|url-status=live}}</ref> Lee apologized for his losses, stating after game three that "I misjudged the capabilities of AlphaGo and felt powerless."<ref name="latimes milestone" /> He emphasized that the defeat was "Lee Se-dol's defeat" and "not a defeat of mankind".<ref name="koreatimes beatable" /><ref name="phys.org challenge" /> Lee said his eventual loss to a machine was "inevitable" but stated that "robots will never understand the beauty of the game the same way that we humans do."<ref name="phys.org challenge" /> Lee called his game four victory a "priceless win that I (would) not exchange for anything."<ref name="koreatimes beatable" />

== AlphaGo documentary film (2016)==
=== Reception ===
{{Anchor|Critical response}}
On Rotten Tomatoes the documentary has an average rating of 100% from 10 reviews.<ref name="tomatoes">{{Cite web |title=ALPHAGO |url=https://www.rottentomatoes.com/m/alphago |website=Rotten Tomatoes |access-date=15 April 2023 }}</ref> <!-- Metacritic has 4 reviews, no critic score listed.<ref>{{Cite web |title=AlphaGo 2017 |url=https://www.metacritic.com/movie/alphago |website=metacritic |access-date=15 April 2023 |archive-date=28 March 2023 |archive-url=https://web.archive.org/web/20230328154537/https://www.metacritic.com/movie/alphago |url-status=live }}</ref> -->

Michael Rechtshaffen of the {{wipe|Los Angeles Times}} gave the documentary a positive review and said: "It helps matters when you have a group of engaging human subjects like soft-spoken Sedol, who's as intensively contemplative as the game itself, contrasted by the spirited, personable Fan Hui, the Paris-based European champ who accepts an offer to serve as an advisor for the DeepMind team after suffering a demoralizing AI trouncing". He also mentioned that with the passion of Hauschka's Volker Bertelmann, the film's producer, this documentary shows many unexpected sequences, including strategic and philosophical components.<ref>{{Cite web |last=Rechtshaffen |first=Michael |date=October 26, 2017 |title=Review: Ancient Chinese board game treated with NFL-like drama and intrigue in documentary 'AlphaGo' |url=https://www.latimes.com/entertainment/movies/la-et-mn-capsule-alpha-go-review-20171026-story.html |website=www.latimes.com |access-date=15 April 2023 |archive-date=15 April 2023 |archive-url=https://web.archive.org/web/20230415175127/https://www.latimes.com/entertainment/movies/la-et-mn-capsule-alpha-go-review-20171026-story.html |url-status=live }}</ref> (Rechtshaffen, 2017
John Defore of The Hollywood Reporter, wrote this documentary is "an involving sports-rivalry doc with an AI twist." "In the end, observers wonder if AlphaGo's odd variety of intuition might not kill Go as an intellectual pursuit but shift its course, forcing the game's scholars to consider it from new angles. So maybe it isn't time to welcome our computer overlords, and won't be for a while - maybe they'll teach us to be better thinkers before turning us into their slaves."<ref>{{Cite web |last=Defore |first=John |date=September 29, 2017 |title='AlphaGo': Film Review |url=https://www.hollywoodreporter.com/movies/movie-reviews/alphago-1044560/ |website=The Hollywood Reporter |access-date=15 April 2023 |archive-date=13 February 2023 |archive-url=https://web.archive.org/web/20230213152947/https://www.hollywoodreporter.com/movies/movie-reviews/alphago-1044560/ |url-status=live }}</ref>

Greg Kohs, the director of the film, said "The complexity of the game of Go, combined with the technical depth of an emerging technology like artificial intelligence seemed like it might create an insurmountable barrier for a film like this. The fact that I was so innocently unaware of Go and AlphaGo actually proved to be beneficial. It allowed me to approach the action and interviews with pure curiosity, the kind that helps make any subject matter emotionally accessible." Kohs also said that "Unlike the film's human characters – who turn their curious quest for knowledge into an epic spectacle with great existential implications, who dare to risk their reputation and pride to contest that curiosity – AI might not yet possess the ability to empathize. But it can teach us profound things about our humanness – the way we play board games, the way we think and feel and grow. It's a deep, vast premise, but my hope is, by sharing it, we can discover something within ourselves we never saw before".<ref>{{Cite web |last=Kohs |first=Greg |date=October 23, 2018 |title=Five Questions for Filmmakers: AlphaGo |url=https://www.sciencemediasummit.org/blog/five-questions-for-filmmakers-alphago |website=Science Media Awards & Summit in the Hub (SMASH) |access-date=15 April 2023 |archive-date=28 March 2023 |archive-url=https://web.archive.org/web/20230328154832/https://www.sciencemediasummit.org/blog/five-questions-for-filmmakers-alphago |url-status=live }}</ref>

=== Professional Go player ===
Hajin Lee, a former professional Go player, described this documentary as being "beautifully filmed". In addition to the story itself, the feelings and atmosphere were also conveyed through different scene arrangements. For example, the close-up shots of Lee Sedol when he realizes that the AlphaGo AI is intelligent, the atmospheric scene of the Korean commentator's distress and affliction following the first defeat, and the tension being held inside the room. The documentary also tells a story by describing the background of AlphaGo technology and the customs of the Korean Go community. She suggests some areas to be covered additionally. For instance, the details of the AI prior to AlphaGo, the confidence and pride of the professional Go players, and the shifting of perspective to the Go AI between and after the match as "If anything could be added, I would include information about the primitive level of top Go A.I.s before AlphaGo, and more about professional Go players' lives and pride, to provide more context for Lee Sedol's pre-match confidence, and Go players' changing perception of AlphaGo as the match advanced".<ref>{{Cite web |last=Lee |first=Hajin |date=Apr 28, 2017 |title=AlphaGo" Film Review: The Art of Capturing the Essence |url=https://hajinlee.medium.com/alphago-film-review-the-art-of-capturing-the-essence-892b70d33e92 |website=hajinlee.medium.com |access-date=15 April 2023 |archive-date=15 April 2023 |archive-url=https://web.archive.org/web/20230415173039/https://hajinlee.medium.com/alphago-film-review-the-art-of-capturing-the-essence-892b70d33e92 |url-status=live }}</ref>(Lee, 2017).

Fan Hui, a professional Go player, and former player with AlphaGo said that "DeepMind had trained AlphaGo by showing it many strong amateur games of Go to develop its understanding of how a human plays before challenging it to play versions of itself thousands of times, a novel form of reinforcement learning which had given it the ability to rival an expert human. History had been made, and centuries of received learning overturned in the process. The program was free to learn the game for itself.<ref name=":1">{{Cite web |last=Williams |first=Rhiannon |date=October 8, 2020 |title=Fan Hui: What I learned from losing to DeepMind's AlphaGo |url=https://inews.co.uk/news/technology/fan-hui-what-i-learned-from-losing-to-deepminds-alphago-google-295005 |website=inews.co.uk |access-date=15 April 2023 |archive-date=28 March 2023 |archive-url=https://web.archive.org/web/20230328154538/https://inews.co.uk/news/technology/fan-hui-what-i-learned-from-losing-to-deepminds-alphago-google-295005 |url-status=live }}</ref>

=== Technology and AI-related fields ===
James Vincent, a reporter from The Verge, comments that "It prods and pokes viewers with unsubtle emotional cues, like a reality TV show would. "Now, you should be nervous; now you should feel relieved". The AlphaGo footage slowly captures the moment when Lee Sedol acknowledges the true power of AlphaGo AI. In the first game, he had more experience than his human-programmed AI, so he thought it would be easy to beat the AI. However, the early game dynamics were not what he expected. After losing the first match, he became more nervous and lost confidence. Afterward, he reacted to attacks by saying that he just wanted to win the match, unintentionally displaying his anger, and acting in an unusual way. Also, he spends 12 minutes on one move, while AlphaGo only takes a minute and a half to respond. AlphaGo weighs each alternative equally and consistently. No reaction to Lee's fight. Instead, the game continues as if he was not there.

James also said that "suffice to say that humanity does land at least one blow on the machines, through Lee's so-called "divine move". "More likely, the forces of automation we'll face will be impersonal and incomprehensible. They'll come in the form of star ratings we can't object to, and algorithms we can't fully understand. Dealing with the problems of AI will take a perspective that looks beyond individual battles. AlphaGo is worth seeing because it raises these questions" <ref>{{Cite web |last=Vincent |first=James |date=October 12, 2017 |title=How will we face being defeated by machines? |url=https://www.theverge.com/2017/10/11/16460118/alphago-deepmind-ai-documentary-go-lee-sedol-film-review |website=www.theverge.com |access-date=15 April 2023 |archive-date=15 April 2023 |archive-url=https://web.archive.org/web/20230415173038/https://www.theverge.com/2017/10/11/16460118/alphago-deepmind-ai-documentary-go-lee-sedol-film-review |url-status=live }}</ref>(Vincent, 2017)

Murray Shanahan, a professor of cognitive robotics at Imperial College London, critics that "Go is an extraordinary game but it represents what we can do with AI in all kinds of other spheres," says Murray Shanahan, professor of cognitive robotics at Imperial College London and senior research scientist at DeepMind, says. "In just the same way there are all kinds of realms of possibility within Go that have not been discovered, we could never have imagined the potential for discovering drugs and other materials."<ref name=":1" />

=== Spiritual and cultural expert ===
Director Greg Kohs brings out all of the drama and rush of these games to show the selection process for the master of Go. On this small stage, the human Go champion and the AI challenger engage in an intense duel. In this documentary, Kohs looks into how the human mind functions under pressure, the significance of errors, the actions of the computer, and its inventiveness.<ref>{{Cite web |last1=Brussat |first1=Frederic |last2=Brussat |first2=Mary Ann |title=A snappy and mind-boggling documentary about a 3000-year-old board game and a A.I. program. |url=https://www.spiritualityandpractice.com/films/reviews/view/28500/alphago |website=www.spiritualityandpractice.com |access-date=15 April 2023 |archive-date=15 April 2023 |archive-url=https://web.archive.org/web/20230415173038/https://www.spiritualityandpractice.com/films/reviews/view/28500/alphago |url-status=live }}</ref> (Frederic and Mary Ann Brussat, n.d.)

==Similar systems==
[[Company:Facebook|Facebook]] has also been working on its own Go-playing system ''[[Darkforest|darkforest]]'', also based on combining machine learning and [[Monte Carlo tree search]].<ref name=":0" /><ref name=Facebook-paper>{{Cite arXiv|eprint=1511.06410v1|last1=Tian|first1=Yuandong|title=Better Computer Go Player with Neural Network and Long-term Prediction|last2=Zhu|first2=Yan|class=cs.LG|year=2015}}</ref> Although a strong player against other computer Go programs, as of early 2016, it had not yet defeated a professional human player.<ref>{{Cite news|title = No Go: Facebook fails to spoil Google's big AI day|url = https://www.theguardian.com/technology/2016/jan/28/go-playing-facebook-spoil-googles-ai-deepmind|newspaper = The Guardian|date = 28 January 2016|access-date = 1 February 2016|issn = 0261-3077|language = en-GB|last = HAL 90210|archive-date = 15 March 2016|archive-url = https://web.archive.org/web/20160315202426/http://www.theguardian.com/technology/2016/jan/28/go-playing-facebook-spoil-googles-ai-deepmind|url-status = live}}</ref> Darkforest has lost to CrazyStone and Zen and is estimated to be of similar strength to CrazyStone and Zen.<ref>{{cite web|url=http://livestream.com/oxuni/StracheyLectureDrDemisHassabis|title=Strachey Lecture – Dr Demis Hassabis|work=The New Livestream|access-date=17 March 2016|archive-date=16 March 2016|archive-url=https://web.archive.org/web/20160316020203/http://livestream.com/oxuni/StracheyLectureDrDemisHassabis|url-status=live}}</ref>

[[Software:Zen|DeepZenGo]], a system developed with support from video-sharing website Dwango and the University of Tokyo, lost 2–1 in November 2016 to Go master Cho Chikun, who holds the record for the largest number of Go title wins in Japan.<ref>{{cite news|title=Go master Cho wins best-of-three series against Japan-made AI|url=http://www.japantimes.co.jp/news/2016/11/24/national/go-master-cho-wins-best-three-series-japan-made-ai/#.WDtq8x_6zCI|access-date=27 November 2016|work=The Japan Times Online|date=24 November 2016|archive-date=14 August 2017|archive-url=https://web.archive.org/web/20170814022025/http://www.japantimes.co.jp/news/2016/11/24/national/go-master-cho-wins-best-three-series-japan-made-ai/#.WDtq8x_6zCI|url-status=live}}</ref><ref>{{cite news|title=Humans strike back: Korean Go master bests AI in board game bout|url=https://www.cnet.com/news/humans-strike-back-korean-go-master-beats-ai-in-board-game-bout/|access-date=27 November 2016|work=CNET|archive-date=25 November 2016|archive-url=https://web.archive.org/web/20161125140800/https://www.cnet.com/news/humans-strike-back-korean-go-master-beats-ai-in-board-game-bout/|url-status=live}}</ref>

A 2018 paper in Nature cited AlphaGo's approach as the basis for a new means of computing potential pharmaceutical drug molecules.<ref>{{cite web|url=https://www.theengineer.co.uk/go-make-drugs/|title=Go and make some drugs The Engineer|website=www.theengineer.co.uk|date=3 April 2018|language=en-UK|access-date=2018-04-03|archive-date=3 April 2018|archive-url=https://web.archive.org/web/20180403144812/https://www.theengineer.co.uk/go-make-drugs/|url-status=live}}</ref><ref>{{cite journal|journal=Nature|volume=555|pages=604–610|date=March 29, 2018|first1=Martwin H.S.|last1=Segler|first2=Mike|last2=Preuss|first3=Mark P.|last3=Waller|title=Planning chemical syntheses with deep neural networks and symbolic AI|issue=7698|doi=10.1038/nature25978|pmid=29595767|arxiv=1708.04202|bibcode=2018Natur.555..604S|s2cid=205264340|url=https://www.nature.com/articles/nature25978|access-date=12 December 2021|archive-date=12 December 2021|archive-url=https://web.archive.org/web/20211212221242/https://www.nature.com/articles/nature25978|url-status=live}}</ref>

==Example game==

AlphaGo Master (white) v. Tang Weixing (31 December 2016), AlphaGo won by resignation. White 36 was widely praised.

{| style="display:inline; display:inline-table;"
|-
| style="border: solid thin; padding: 2px;" |
{{Goban
|  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |
|  |  |  |  |  |  |  |  |  |  |  |  |  |  |69|  |  |  |
|  |  |  |  |  |  |  |  |  |  |  |  |83|68|  |  |  |  |
|  |  |  |w2|  |  |  |  |  |  |  |  |  |  |  |b1|  |  |
|  |  |  |  |  |  |  |  |  |  |  |  |  |73|  |  |71|  |
|  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |72|70|  |
|  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |
|  |  |58|  |  |  |  |  |  |  |  |  |  |  |  |76|  |  |
|  |  |56|57|  |  |  |  |  |  |  |  |82|78|  |74|75|  |
|  |54|15|  |  |  |  |  |  |  |  |  |80|79|  |77|67|  |
|  |52|53|36|37|  |47|  |  |  |  |  |  |81|  |  |  |  |
|  |48|39|27|  |46|41|40|51|92|  |  |  |  |  |  |84|  |
|  |42|38|b9|  |  |44|45|  |26|  |91|  |  |  |  |98|93|
|64|43|b7|w8|10|  |  |25|  |  |90|89|  |85|  |87|  |  |
|  |  |13|22|  |28|29|19|24|  |  |66|  |86|  |88|w6|99|
|  |62|11|w4|  |21|  |17|18|  |  |  |  |30|  |34|97|96|
|  |49|12|14|  |b5|  |20|16|  |  |  |23|31|35|b3|32|95|
|55|50|59|  |  |65|  |  |  |  |  |  |  |  |  |  |33|94|
|63|60|61|  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |
|20}}
|-
| style="text-align:center" | First 99 moves
|}
{| style="display:inline; display:inline-table;"
|-
| style="border: solid thin; padding: 2px;" |
{{Goban
|  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |
|  |48|46|47|  |36|54|53|51|  |27|14|13|17|b |  |85|  |
|  |34|29|33|35|w8|  |52|26|b9|23|12|b |w |  |  |83|84|
|42|31|30|w |37|  |  |  |  |22|18|15|10|11|  |b |  |80|
|  |32|41|43|68|40|38|70|  |24|21|19|16|b |  |  |b |81|82
|  |45|  |  |44|39|64|63|  |69|  |  |20|71|  |w |w |78|
|  |  |  |  |  |65|66|  |  |  |75|25|72|73|  |  |  |  |
|  |  |w |28|  |67|  |  |  |  |  |  |74|  |  |w |  |  |
|  |  |w |b |  |  |  |  |  |86|  |  |w |w |  |w |b |  |
|  |w |b |  |  |  |  |  |76|55|  |  |w |b |  |b |b |b3|
|  |w |b |w |b |  |b |77|  |62|w6|  |w4|b |  |  |79|w2|
|  |w |b |b |  |w |b |w |b |w |  |56|b5|57|  |  |w |b7|
|  |w |w |b |  |  |w |b |  |w |  |b |  |  |  |  |w |b |
|w |b |b |w |w |  |  |b |  |  |w |b |  |b |  |b |00|  |
|  |  |b |w |  |w |b |b |w |  |  |w |  |w |  |w |w |b |
|  |w |b |w |  |b |  |b |w |  |60|58|  |w |  |w |b |w |b1
|  |b |w |w |  |b |  |w |w |61|59|  |b |b |b |b |  |b |
|b |  |b |  |  |b |  |  |  |  |  |  |  |  |  |  |b |w |
|b |  |b |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |
|20}}
|-
| style="text-align:center" | Moves 100–186 (149 at 131, 150 at 130)
|}

==Impacts on Go==
The documentary film ''AlphaGo''<ref name="alphagomovie"/><ref name="tomatoes" /> raised hopes that Lee Sedol and Fan Hui would have benefitted from their experience of playing AlphaGo, but {{as of|May 2018|lc=on}}, their ratings were little changed; Lee Sedol was ranked 11th in the world, and Fan Hui 545th.<ref>{{cite web |url=https://www.goratings.org/en/ |title=Go Ratings |publisher=Go Ratings |access-date=2018-06-05 |archive-date=15 August 2021 |archive-url=https://web.archive.org/web/20210815082130/https://www.goratings.org/en/ |url-status=live }}</ref> On 19 November 2019, Lee announced his retirement from professional play, arguing that he could never be the top overall player of Go due to the increasing dominance of AI. Lee referred to them as being "an entity that cannot be defeated".<ref>{{Cite web|url=https://www.theverge.com/2019/11/27/20985260/ai-go-alphago-lee-se-dol-retired-deepmind-defeat|title=Former Go champion beaten by DeepMind retires after declaring AI invincible|last=Vincent|first=James|date=2019-11-27|website=The Verge|language=en|access-date=2019-11-28|archive-date=7 April 2020|archive-url=https://web.archive.org/web/20200407012429/https://www.theverge.com/2019/11/27/20985260/ai-go-alphago-lee-se-dol-retired-deepmind-defeat|url-status=live}}</ref>

==See also==
{{div col|colwidth=30em}}
* [[Biography:Albert Lindsey Zobrist|Albert Lindsey Zobrist]], wrote first Go program in 1968
* Chinook (draughts player), draughts playing program
* [[Deep reinforcement learning]], subfield of machine learning that is the basis of AlphaGo
* [[Glossary of artificial intelligence]]
* [[Go and mathematics]]
* [[Software:Leela|Leela]]
* [[Software:Leela Zero|Leela Zero]], open-source learning Go program
* [[Matchbox Educable Noughts and Crosses Engine]]
* [[Biography:Arthur Samuel (computer scientist)#Computer checkers (draughts) development|Samuel's learning computer checkers (draughts)]]
* [[TD-Gammon]], backgammon neural network
* Pluribus (poker bot)
* [[AlphaZero]]
* [[AlphaFold]]
{{div col end}}

==References==
{{Reflist}}

==External links==
* [http://senseis.xmp.net/?AlphaGo AlphaGo wiki] at [[Software:Sensei's Library|Sensei's Library]], including links to AlphaGo games
* [http://homepages.cwi.nl/~aeb/go/games/games/AlphaGo/ AlphaGo page], with archive and games
* [https://web.archive.org/web/20170104201657/https://www.goratings.org/ Estimated 2017 rating of Alpha Go]

{{Differentiable computing}}


[[Category:Applied machine learning]]

{{Sourceattribution|AlphaGo|1}}