A few weeks back, we had the pleasure of hosting the Silicon Valley Artificial Intelligence Meetup group at our office here in San Francisco. About 100 folks showed up to learn how Google’s DeepMind AlphaGo program beat professional Go player Lee Sedol. Our expert for the evening was Robert A. Hearn, who is both an MIT PhD in Computational Neuroscience and an avid Go player himself. It was a delight.
What we enjoyed most about the talk was its depth. Most of the stories that have come out in the month plus since the historic matches haven’t gotten into the meat and potatoes of the achievement. Yes, they note it was a historic win, but beyond explaining that the victory was unexpected, they didn’t often delve into the tech behind AlphaGo, how it works and actually makes decisions, or how quickly the state-of-the-art evolved from “fairly miserable” to “just beat a world champion.”
Robert Hearn, however, did:
Robert talks about everything from the rules of Go to the specifics behind AlphaGo’s policy and value networks to the moves that won the games themselves. The talk is an insightful look into what an immense challenge Go posed for artificial intelligence and how, in the end, it was actually solved.
We’ll be doing some more good stuff with the Silicon Valley AI Meetup in the coming months and, of course, we’ll keep you posted.