Three Games Artificial Intelligence Seems To Be Winning

The idea of artificial intelligence overcoming humans has long been the subject of science fiction. Films like Metropolis (1927) and Terminator (1984) seemed removed from reality when released, yet in the 21st century, certain developments could easily render this sort of science fiction into fact.

Though AI has yet to irrevocably pass the infamous Turing test, one marker of its development has always remained constant: defeating man in games of logic and skill.

Chess

In 1965, MIT professor, Herbert Dreyfus, declared in the book, What Computers Can Do, that “no computer could beat a 10-year-old at chess”. Two years later, Dreyfus lost a “real cliff-hanger” of a match against Mack Hack VI, a computer designed by students at his university.

Despite this, there were doubts over AI’s ability to beat the best chess players in the world. Yet that is exactly what happened in 1981. That year, the computer Cray Blitz took part in the Mississippi State Championship and scored 5-0. It also beat chess master, Joe Sentef, in the fourth round of the competition.

Deep Blue’s defeat of Garry Kasparov in 1996 marked the first time that a computer beat a reigning world champion. Yet some decried this as void due to an alleged glitch in the system. The doubters felt vindicated when Kasparov beat Deep Blue in a rematch the following year.

However, since then, several masters have lost against computers. A program running on an HTC mobile phone even reached grandmaster level in 2009. It seems where a tactical game of chess is concerned; both sides are in stalemate.

Go

[pullquote]With artificial intelligence now reigning victorious over humans in the chess world, it was time to test its mettle at more complex games.[/pullquote] The 2500-year-old game Go certainly fit the bill.

This ancient game, which finds its origins in China, sets two players against each other on a grid. The aim is to capture more territory than the opponent by placing stones on this grid – seemingly innocuous enough. However, there are actually more combinations available than there are atoms in the visible universe, hinting at the game’s level of complexity.

Up until 2015, computers playing against human players of the game could only really match at an amateur level. That all changed in January 2016, when Google’s own AlphaGo trounced reigning European champion, Fan Hui, in a game that ended 5-0.

Later that year, AlphaGo faced 18-time Go world champion Lee Sedol. Again, AlphaGo dominated the competition – donating the entire million-dollar prize fund to charity.

Poker

In May 2015, TechTimes reported that humans beat a computer at poker in over 80,000 matches. This was unsurprising considering that poker is more difficult for computers to grasp as a result of its “imperfect information”. This arises from the fact that in chess and Go, a computer can see the whole board.

By contrast, the computer may not see its opponents’ hands in poker. So, while the computer is by no means stupid, it still surely remains possible to beat. This is, of course, why a whole host of online gaming sites – such as onlineroulette.org.uk – remain perennially popular with players, even if they are placing wagers against a piece of software or algorithm.

Yet in January 2017, news came that computer ‘Liberatus,’ beat four of the best players in the world at poker. The 20-day tournament saw Dong Kim, Jason Les, Jimmy Chou and Daniel McAulay playing against the computer for 11 hours a day.

By the end of the competition, Liberatus had won more than $1.7million in poker chips from the four players. However, worth more than the money was the critical claim its creators could now make: another machine had outmaneuvered mankind. This proved a watershed moment for artificial intelligence and brings home just how powerful computers may become in the not-too-distant future.