Lee Se-dol Retaliates And Beats Google’s AI For The First Time

DeepMind’s AlphaGo artificial intelligence program was leading in a series of five matches against human Go champion Lee Se-dol until the latter retaliated in a surprising move for Google’s machine learning player yesterday.

After three consecutive loses, Se-dol took his revenge striking with move 78, forcing AlphaGo to follow with some disastruos moves. Demis Hassabis, DeepMind’s founder, kept its objectivity throughout tweeting like a good sport:

 

Lee Sedol wins game 4!!! Congratulations! He was too good for us today and pressured #AlphaGo into a mistake that it couldn’t recover from

— Demis Hassabis (@demishassabis) March 13, 2016

While AlphaGo can simulate millions of games of Go, “seeing” the outcomes and continously trying to build a general strategy, experienced players don’t benefit only from the matches they’ve witnessed and learned from, but will come up with ingenous, original moves like Se-dol did. Since there are more possible positions in Go than the number of atoms in the universe, a genius human player can have the upperhand multiple times.

Another enduring drawback of AI is the fact that they are programmed for one specific task and are unable, unlike humans, to operate in a chaotic, complex world like the one of our society.