As Google’s DeepMind team bathed in the accolades that came with building an artificial intelligence machine that decisively kicked the butt of one of the world’s best human Go players, it seems jealousy reared its ugly head.
On Monday, DeepMind’s AlphaGo program defeated Lee Sedol, an 18-time Go champion, in the final match of the best-of-five tournament. The triumph by AlphaGo, which won the series 4-1, was the first time an AI machine has defeated a human champion in the ancient Chinese game. The victory was widely celebrated in tech and AI circles — save for one notable voice.
Yann LeCun, Facebook’s director of artificial intelligence research, started out well. He took to the social network to congratulate Google, but that’s when he apparently ran out of graciousness.
After AlphaGo recorded a third consecutive triumph over Sedol, LeCun wrote: “Congrats to the DeepMind AlphaGo team for this Grand Slam. Now, can you do it purely through reinforcement learning, without pre-training the convolutional net on recorded games between humans?”
LeCun is apparently skeptical about whether AlphaGo is actually learning how to play, or is simply processing millions of potential Go moves that have been programmed into its memory. After Sedol recorded his only win on Sunday, LeCun wrote: “let’s not discount the power of human learning too quickly,” and “the successor of AlphaGo should be called BettaGo.” He linked to coverage critical of AlphaGo’s victory and suggested the media attention was “overblown.”
Watching Google’s AI efforts soak up all the attention had to be hard for LeCun. In January, Facebook announced that the company’s AI team was close to building a machine that could play Go like a human. And then before Facebook could demonstrate anything, AlphaGo stole all the thunder.
That’s tough, but making petty statements won’t help Facebook in the race to develop a better AI machine. Bashing AlphaGo won’t make Facebook’s AI one bit smarter. A good public-relations manager would have helped LeCun see that before he made himself look spiteful (truth is PR execs love when managers make goofs of themselves because it reminds everybody why they’re needed).
It’s important to note that Facebook CEO Mark Zuckerberg took the high road and wrote: “Congrats to the Google DeepMind team on this historic milestone in AI research … We live in exciting times.”
Maybe Zuck should remind LeCun that the best way to show up Google and AlphaGo is to build something better.