The fact that computers are getting increasingly smarter is evident. However, they still lack creativity and intuition and hence cannot learn, think and understand like humans — At least, that was the consensus until Google's AlphaGo computer algorithm beat 18-time world champion Lee Se-dol, 4 to 1 at the ancient Chinese game, Go.
To understand why that is a big deal one has to know the intricacies of the 3,000-year-old board game that is still very popular in Asia. Similar to chess, Go features a checkered grid and black and white game pieces called "stones." The two players take turns placing the stones on the intersections of a grid. Their aim is to cover as much of the grid as possible by sealing off areas and surrounding each other's stones. Once that happens, the opponent's stones are removed from the board. The player with the most stones remaining on the grid at the end is declared the winner.
Though that sounds simple enough thanks to its 19x19 grid, there are an inordinate number of ways the board can be configured — Over 10761, or more than all the atoms in the universe to be exact! It was believed that a machine could not anticipate all the possible outcomes and therefore, never be able to outdo a human.
While that perception changed somewhat in October 2015, when AlphaGo beat the European Go Champion Fan Hui 5-0, most people still believed that Lee Se-dol would make short work of the computer algorithm. The South Korean world champion who has 18 titles under his belt is after all, considered the Go grandmaster.
The epic face-off between the two that was staged in Seoul, South Korea from March 9 - 15, drew tremendous public interest, with tens of millions of people in China and other countries tuning in to watch the live streams. The match even intrigued people in the US, where Go is relatively unknown.
AlphaGo's decisive win in four of the five games stunned everyone including Google's DeepMind team that developed the program. DeepMind founder Demis Hassabis attributes AlphaGo's success to its two “deep neural networks” that allow it to search through a database containing information on millions of moves made by the world's best human Go players. It also 'rewards' the program for every right move, keeping it motivated to win. To prepare for the challenge, the algorithm played the game with itself millions of times. This helped the programmers fine tune the strategies that worked the best.
All this hard work was handsomely rewarded. In addition to the prize money of $1 million USD (which was donated to charity), the algorithm was also awarded an honorary 'ninth dan' - the highest possible Go rank by the South Korean Go Association.
While experts worldwide are impressed at the algorithm's quick thinking powers, their opinions about it are mixed. Some like Tesla founder Elon Musk and Apple co-founder Steve Wozniak, view the technological advancement as dangerous. Others believe that we have nothing to worry about. They think that AlphaGo lacks compassion and is only able to process information on a game-by-game basis, which means unlike humans, it is unable to see the bigger picture.
AlphaGo is also programmed to have efficient time management, meaning it may choose the fastest option instead of taking longer and utilizing its best judgment. That means that while you may not want to compete against the software in a video or board game, there is still a while to go before the computer starts to replace humans.
Besides, Lee is not ready to accept defeat just yet. The Go grandmaster who later apologized to his fans for not taking the computer program seriously, believes he now knows how its 'mind' operates and is ready for another challenge — So stay tuned for the next epic battle between man and machine!
Resources: theguardian.com,csmonitor.com,independent.co.uk, fortune.com