AI software could now beat humans at chess after reading the rule book
Google’s Artificial Intelligence (AI) can now teach itself to beat humans at complex games without using knowledge given to it by human developers.
AlphaZero, the game-playing AI created by Google sibling DeepMind, is able to master games like chess, shogi and Go just by reading the rule book.
The ‘ superhuman’ computer program teaches itself to play these games with no prior knowledge except each game’s rules.
A study, led by American Association for the Advancement of science, shows that the program was able to teach itself the intricacies of each game until mastered.
The creators of the software say it marks a ‘turning point’ for intelligent machines that can think more creatively and intuit like people.
An earlier version of the machine, dubbed AlphaGo, was able to defeat the world’s top human players of the Chinese board game Go.
Since IBM’s chess program, Deep Blue beat champion Garry Kasparov in 1997, game- playing AIs have grown more and more advanced.
Based on self-play reinforcement learn- ing, David Silver and colleagues at DeepMind developed AlphaZero, a generalised game-playing program that foregoes the need for human-derived information.
The system has become increasingly able to beat humans at highly complex games such as shogi and Go, each significantly more difficult than chess.
However, the algorithms that drive these AI systems are often made to exploit the properties of a single game and rely on knowledge they are programmed with by their human developers, according to the authors of the study.
The researchers found that AlphaZero was able to learn chess, shogi and Go by playing against itself - repeatedly - until each was mastered.
According the team, the system was able to beat state-of-the-art AI programs which specialised in these three games after just a few hours of self-learning.
Speaking to The Telegraph, Professor David Silver of DeepMind said: ‘It’s got a very subtle sense of intuition which helps it balance out all the different factors.
‘It’s got a neural network with millions of different tunable parameters, each learning its own rules of what is good in chess, and when you put them all together you have something that expresses, in quite a brain-like way, our human ability to glance at a position and say “ah ha this is the right thing to do”.
‘ My personal belief is that we’ve seen something of turning point where we’re starting to understand that many abilities, like intuition and creativity, that we previously thought were in the domain only of the human mind are actually accessible to machine intelligence as well. And I think that’s a really exciting moment in history.’
Despite the immense complexity of games like chess, shogi and Go, recent advancements in AI have rendered them into easily solvable problems.
As a result, AI researchers need to look to a new generation of games - multiplayer video games, for example - to provide the next set of challenges for AI systems.
A study of AlphaZero last year showed that it defeated one of the best chess programs in the world after learning the game from scratch in just four hours.
The AI played 100 games against rival computer program Stockfish 8, and won or drew all of them.