Last week the big news was that Google’s AlphaGo was able to win 4 of the 5 games against Lee Sedol in Go. As we’ve gotten better and better hardware it’s not surprising that an AI was finally able to win in a well defined environment. AIs will continue to improve and we’ll start seeing more and more of this behavior across a wide range of problems and not just games. The most significant part for me was that this was achieved by Google and not by IBM. IBM had two recent notable achievements in AI - one was building Deep Blue in 1997 which beat Gary Kasparov in chess and the other was building Watson in 2001 which dominated at Jeopardy. Yet just five years later Google has claimed the AI victory with AlphaGo.
It’s tough to not see this is as the passing of the torch. Google has tons of incredible AI that’s used to power the business but building an AI to focus on Go was taking something out of IBM’s playbook. Luckily, we’re seeing a ton of AI work being open sourced - from Google’s TensorFlow to Facebook’s FAIR to OpenAI - and I’m excited to see what we come up with.