AlphaGo

AlphaGo versus Lee Sedol, known as the DeepMind Challenge Match, represents a landmark moment in artificial intelligence history, comparable to the 1997 chess match when Deep Blue defeated Garry Kasparov. This match demonstrated that AI had conquered one of humanity's most complex and ancient games, which experts believed was at least a decade away from being solved by computers.

The match took place March 9-15, 2016 at the Four Seasons Hotel in Seoul, South Korea. AlphaGo won 4-1 against Lee Sedol, with the prize of $1 million (donated to UNICEF and Go organizations). Go is a complex board game requiring intuition, creative and strategic thinking. It was considered considerably more difficult to program than chess because it requires more elements that mimic human thought. Mathematician I. J. Good wrote in 1965 that programming a computer to play Go would be even more difficult than chess because 'the principles are more qualitative and mysterious than in chess, and depend more on judgement.' Prior to 2015, the best Go programs only managed to reach amateur dan level.

AlphaGo is significantly different from previous AI efforts. Instead of using probability algorithms hard-coded by human programmers, AlphaGo uses neural networks to estimate its probability of winning. The system employs machine learning and tree search techniques combined with extensive training. Neural networks were initially bootstrapped from human game-play expertise, training from a KGS Go Server database of around 30 million moves from 160,000 games by 6-9 dan human players. It used reinforcement learning through playing large numbers of games against itself, and Monte Carlo tree search to calculate colossal numbers of probabilities many moves into the future. The system used 1,202 CPUs and 176 GPUs (some sources report 1,920 CPUs and 280 GPUs), with Google's proprietary tensor processing units (TPUs) employed in the match. The server was located in the United States, accessed through Google Cloud Platform.

The match generated massive global media coverage and public interest in AI. Lee Sedol's emotional response humanized the AI challenge. It sparked philosophical debates about human vs. machine intelligence and demonstrated AI's potential to surpass human expertise in domains requiring intuition and creativity. Game 1 saw AlphaGo make one unusual move that no human Go player would have made, demonstrating creative play beyond human intuition. In Game 4, Lee Sedol's only victory, he earned an additional $20,000 (total $170,000). His 'Move 78' was described as brilliant and unexpected, showing human creativity could still surprise AI.

Before the match, experts thought AI was 10 years away from achieving victory against a top professional Go player. Elon Musk, an early investor in DeepMind, said in 2016 that this timeline represented the expert consensus. Lee Sedol initially predicted he would defeat AlphaGo in a 'landslide.' The Korea Baduk Association awarded AlphaGo the highest Go grandmaster rank: honorary 9 dan, in recognition of AlphaGo's 'sincere efforts' to master Go. The achievement was chosen by Science magazine as one of the runners-up for Breakthrough of the Year on December 22, 2016.

The match demonstrated that neural networks and reinforcement learning could solve problems previously thought impossible. Research results were applied to fields such as cognitive science, pattern recognition, and machine learning. It inspired further development including AlphaGo Zero (2017), which learned entirely through self-play without human game data, and showed AI could develop intuition and creativity, not just brute-force calculation. Prior to AlphaGo's victory, in October 2015, AlphaGo defeated European champion Fan Hui, a 2 dan professional, 5-0, marking the first time an AI had beaten a human professional player on a full-sized board without handicap. Fan Hui noted that playing against AlphaGo taught him to be a better player, and his world ranking rose from 633 to around 300 by March 2016.