The Imitation Game
How is it that we continue to underestimate the intelligence of AI, despite its many accomplishments? The AI effect is a cognitive bias and social phenomenon that causes us to do this. Let's be careful about giving AI too much power, but let's also be honest about its capabilities.
When the chess computer Deep Blue beat Garry Kasparov in 1997, many people were surprised. They had expected Kasparov, one of the greatest chess players in history, to win easily. Computers were not supposed to be able to beat humans at chess, and yet Deep Blue had done it.
Deep Blue's victory was a milestone in the development of artificial intelligence. It showed that computers could be programmed to play chess at a high level. Since then, there have been many other milestones in AI. In 2016, an AI program called AlphaGo beat a Go champion, Lee Sedol, in a five-game match. This was even more surprising than Deep Blue's victory, because Go is a much more complex game than chess.
In the past few years, AI has made tremendous progress. There are now AI programs that can drive cars, diagnose diseases, and trade stocks. AI is transforming many industries, and it is having a big impact on our lives. Yet, we seem to be looking for ways to downplay how intelligent AI has become. Do we still consider a chess computer to be intelligent, even though it can beat the best human players? Or do we say that it's just a chess machine? The moment we are getting used to a certain piece of technology, we somehow stop calling it AI.