Rise of the machines: when a computer beats a chess master
While computational supremacy of digital devices seems like a given in 2022, the debate was far from settled not so long ago.
A quarter of a century ago, on 11 May 1997, a compelling argument for computer supremacy was made. For the first time in history, a computer defeated a world chess champion in a match of several games.
The IBM-made device dubbed 'Deep Blue' defeated the reigning champion, Garry Kasparov, in an unusually swift chess game. The sixth game lasted only 19 moves, while an average game takes around 40 to complete.
In the 1997 match, Kasparov had won the first game, lost the second, and drawn three subsequent games. The loss of the sixth match marked a clear defeat of a human chess master by an artificial machine. A feat that has never occurred in over 1,500 years of chess history.
Years in the making
Kasparov's defeat did not fall out of the blue. The computer that eventually outsmarted humanity has been in the works since 1985 when a Taiwanese American computer scientist Feng-hsiung Hsu set out to build a machine that could beat humans in chess.
The first purpose-built chess computer was named Deep Thought, after a fictional computer in the well-known novel 'The Hitchhiker's Guide to the Galaxy.' Kasparov took onto the device as early as 1989, but the chess master easily defeated the computer in a two-game match.
A publicity side-project of IBM, the chess computer was in the works for several years, eventually dropping its name in favor of a more IBM-friendly 'Deep Blue.' At the time, IBM was also frequently referred to as the 'Big Blue.'
The first lost game
The second time Deep Blue challenged Kasparov came in 1996. Surprisingly, the computer beat the reigning world champion, the first time a computer won a single game in a chess match.
However, Kasparov defeated the computer in the second, fifth, and sixth games and agreed to a draw in the third and fourth games, thus claiming victory in the entire match.
Even though Kasparov was victorious, the 'man vs. machine' narrative caught the attention of the world press, and both parties agreed to another match. While it was credited as a rematch, Kasparov did not face the same Deep Blue computer he did in 1996.
According to IBM, the updated machine was capable of exploring up to 200 million possible chess positions per second. For context, typically, chess tournament rules allow three minutes to make a move. The computer was also updated between every game to compensate for its flaws against Kasparov.
Kasparov lost the much-hyped match in front of a live audience of 500 people and millions who tuned in to see whether humans can still outperform their creations.
Tradition of defeat
Deep Blue was more than a purely for-fun venture. IBM claims that the research enabled the company to explore the limits of massively parallel processing. IBM applied the findings to tackle problems in financial modeling, risk analysis, data mining, and molecular dynamics.
While Deep Blue was retired to the Smithsonian Museum, its success inspired the company to develop Watson, a computer that beat the champions of the game 'Jeopardy!' in 2011. Watson's victory proved that machines are capable of processing natural human language.
The tradition of computers beating professionals at their game never went away. In 2015, DeepMind's computer program AlphaGo defeated a professional player of the immensely complicated game of Go. Updated versions of the computer beat the world's best players in the following years.
The defeat signaled a milestone for the development of Artificial Intelligence (AI) and machine learning since the game of Go is considered a lot more complicated. Pure brute force logic that allowed to defeat Kasparov does not apply in Go.
According to Google, the search space, the feasible region defining the set of all possible solutions, in Go is googol times larger than chess. Googol represents a number that is greater than the number of atoms in the universe. To comprehend a task of this magnitude, predictive neural networks, computer systems that mimic biological brains, were employed.
More from Cybernews:
Subscribe to our newsletter