AlphaGo, a Google-developed artificial intelligence program, has just recently defeated a world champion Go player in two consecutive matches. Murray Campbell, an IBM research scientist who is also part of the team who developed the first computer program (called Deep Blue) to have ever beaten a world chess champion, stated that the era of testing out AI in board games has come to an end. Therefore, it begs the question, what’s next for AI testing?
After AlphaGo Defeats World Champion Go Player, What’s Next?
AlphaGo is now known to be one of the AIs who can beat world champion board gamers. Artificial intelligence is now also known to beat many human beings in IQ tests, general knowledge quizzes, and even SATs. But now that these programs are able to beat specific aspects of the human intellect, what would be next for artificial intelligence?
Campbell stated, “Games are fun and they’re easy to measure.” He then continues, “It’s clear who won and who lost, and you always have the human benchmark. Can you do better than a human?” At the time of writing, the answer to that question is a big, fat “Yes!” Computer algorithms are known to challenge world-champions since the nineties. They have been known to challenge world-class leaders in checkers and in chess. Now, it would seem that another champion is in the history of board games with the win of the AlphaGo AI.
Artificial intelligence is now known to conquer what experts call “complete information games.” These are the kinds of games wherein players can see what their opponents are doing. Tuomas Sandholm, a Carnegie Mellon University professor, studies AI and stated that the next step for artificial intelligence is for “incomplete information games,” such as poker.
The games to be challenged by AI is typically chosen for specific challenges in which the program’s will try and overcome. Board games such as chess and Go already have computer programs that have been put to the test, and have actually won against human opponents. However, these programs were not just made in under a day, they most certainly did not immediately beat world-class human opponents easily.
Also, computer algorithms such as AlphaGo only have one facet of the human mind (so to speak). In other words, it can only handle playing Go and just about nothing else. Therefore, it may still be a good long while before artificial intelligence can be truly part of our human daily lives in more ways than one.