Artificial intelligences have recently bested the world’s top Go and poker players, but those aren’t the first machines to dominate humans in the world of gaming, with systems claiming victories as many as 20 years ago.
Artificial intelligence (AI) is advancing rapidly. From revolutionizing transportation to disrupting the workplace, the technology is poised to completely transform how we live and work in the future, and Elon Musk has claimed that AI will be better than us at everything by 2030.However the future turns out, though, we already have proof that AI can utterly dominate humans within the world of gaming.
AI systems are becoming more and more adept at beating humans at games we have been playing for decades or even centuries. The below are just five such examples of AI’s domination in the world of gaming.
CHESS: IBM’S DEEP BLUE VS. GARRY KASPAROV
When Gary Kasparov, widely considered the greatest chess player in history, faced off with IBM’s chess-playing computer Deep Blue in 1996, he beat the computer 4-2 to the ringing applause of believers in human exceptionalism everywhere.
Despite his win, Kasparov wrote the following in Time soon after the match: “I could feel — I could smell — a new kind of intelligence across the table.” The next year, that new kind of intelligence bested him, with Kasparov falling 3½–2½ in a match against the colloquially named Deeper Blue. The event is widely considered the first time a computer beat a world champion at a game.
BACKGAMMON: BKG 9.8 VS. LUIGI VILLA
In fact, the first time a computer beat a human world champion was around 20 years before. In 1979, Hans J. Berliner watched his backgammon-playing program BKG 9.8 wreck Luigi Villa, the world’s number one player at the time, by a score of 7-1
GO: DEEPMIND’S ALPHAGO VS. FIVE TOP PLAYERS
On to a more recent and multifaceted victory: AlphaGo defeating five of the world’s top players — not one at a time, but simultaneously — at the 2,500-year-old game of Go earlier this year. In the process, AlphaGo showed it was better at the game than multiple humans. The players in the game, though, were not devastated. Instead, they seemed to enjoy adapting to the moves of the computer and learning its style of play.
POKER: LIBRATUS VS. FOUR TOP PLAYERS
In another multi-party encounter, an AI named Libratus convincingly beat four professional poker players at no-limit Texas Hold ‘Em in February. This particular form of poker has no upper limit on betting and gives players the ability to bet all of their chips at any time. The game incorporates bluffing and spontaneity, two traits that seem far more suited for a human player than an AI, and yet the system built by two Carnegie Mellon computer scientists was still able to claim a victory.
Dong Kim, one of the best players in the world, told Wired during the tournament, “I didn’t realize how good it was until today. I felt like I was playing against someone who was cheating, like it could see my cards. I’m not accusing it of cheating. It was just that good.”
CHECKERS: CHINOOK VS. MARION TINSLEY
On a slightly more optimistic note is the competition between an AI called Chinook and former 37-year-straight checkers world champion Marion Tinsley. In their first encounter in 1992, Tinsley won, finding the single strategy that could have defeated Chinook from a point 64 moves ahead. In their second series, in 1994, after drawing six games, Tinsley had to retire due to pancreatic cancer.
Chinook beat the world’s second best player, Don Lafferty, who was still leagues below Tinsley, and so the AI was awarded the win. While Tinsley’s condition was tragic, it means that the AI did not truly beat the world’s best checker player. Later, in 2007, Jonathan Schaeffer, a computer-games expert from the University of Alberta, updated Chinook to the level of unbeatable by finding a perfect way to play the game that would always result in a draw.
THE FUTURE OF AI GAMERS?
AI is getting stronger and stronger at all types of games, but members of the gaming community will truly turn their heads when it is capable of beating humans at StarCraft II. The game is considered the ultimate benchmark due to its complexity. Oriol Vinyals, a Deepmind researcher and former top player, told The Verge that an AI win would have implications beyond the game: “The skills required for an agent to progress through the environment and play StarCraft well could ultimately transfer to real-world tasks.”
While StarCraft II is the current white whale for AI, the victories detailed above are a testament to the technology’s rapid development, and the systems are getting exceeding better at more intuitive and traditionally human activities like composing music and surrealist art. Thankfully, we still have the option of responding the same way Kurt Russell’s character did in The Thing when faced with an unbeatable AI gaming opponent: