One of the most notable events this week has been something that has reminded me of my first ever post on this blog. This has of course been the computer program AlphaGo’s victory at the ancient Chinese boardgame of Wei-qi (as a proud Sinophile who has lived in Taiwan for two years, I’ve spent most of the week being frustrated to the point of rage by everyone describing it as a Japanese game, even though the only differences between Go and Wei-qi are flat bottomed stones versus rounded ones, a slightly different scoring system and a more easily pronounced name) over Lee Sedol; the second highest ranked human player in the world. This has been interpreted by everyone as one more step towards the day when artificial intelligence overtakes human intelligence and all the predictions of the Terminator film series and Battlestar Galactica finally come true. Other recent evidence has been Boston Dynamics video of their new Atlas robot, which can open doors, hike cross-country in the snow and carry and stack boxes, even when a human with a hockey stick is pushing the boxes out of reach, knocking them out of its hands or pushing the robot itself over. The Youtube comments section for the video is filled with people saying variations on two main themes; “That’s me out of work” and “When the robot uprising happens, we are going to be so sorry we made this video.”
A robot rebellion has always seemed like a worryingly real possibility to many sci-fi authors. Creating machines with a higher capabilities than ourselves could turn out to be a mistake if they decide to turn on us. However, I am not convinced that it would doom us to extinction. It is easy to imagine machines with higher capabilities, but this implies that there is a limit to human intelligence; a glass ceiling that we cannot rise above. This might in itself be true, but if so where is it? How do we measure to find it? Intelligence is a very broad term that covers a huge range of skills and abilities, so is the limit the same for all of them or different for each one? Even if we can find the envelope of human intelligence, how do we know for sure that we can’t push it?
Some evidence of this can be found in the game of Centaur Chess (also known as Cyborg or Advanced Chess) a chess variant that Gary Kasparov (not the first man to be defeated at chess by a computer, but the first to have a large international audience there to pour the salt in) came up with after winning several rematches against IBM’s Deep Blue in the 1990s. In Centaur Chess, each human player is allowed to have a computer on hand with a chess-playing program that they can use to calculate possible future moves throughout the game. Kasparov wanted to see what would happen if humans and computers co-operated to play chess instead of competing and the idea has been a great success. Centaur Chess had become very popular with regular international competitions. Wikipedia lists the reasons for its popularity in “increasing the level of play to heights never before seen in chess; producing blunder-free games with the qualities and the beauty of both perfect tactical play and highly meaningful strategic plans and giving the viewing audience an insight into the thought processes of strong human chess players and strong chess computers, and the combination thereof.”
But the most interesting thing about Centaur Chess is that not only do the human players perform better than they can without the computers, the computers perform better than they can on their own. There are numerous theories as to why this is, but my favourite is that human ingenuinty tied to detailed computer analysis is a winning combination. Viswanathan Anand, who is currently the world’s best Centaur Chess player (or at least half of it) has said “I think in general people tend to overestimate the importance of the computer in the competitions. You can do a lot of things with the computer but you still have to play good chess.”
The lesson to take from this is that one of the standard recipes for surviving a robot uprising suggested in science fiction – immediately ditching your smartphone, smashing your self-aware toaster before it can toast you and going off the grid and resorting to non-digital technology to survive – may not be necessary. If instead we can find or build computers and machines that can co-operate with us against their hostile cousins, the advantage is still likely to be ours. Technology does have a tendency to advance both faster and slower than we expect, but I don’t think we need to fear the robot uprising just yet.