Was already excited to read that Facebook have a 5th amateur dan Go bot yesterday, then today Deepmind reveal theirs has beaten a 2nd pro dan! Very cool and can't wait to see their challenge against one of world's best Go players Lee Sedol in March (9th pro dan).

I imagine that even if given access to the source / neural net weights, it would be much harder for a player to plan against this type of Go engine than old chess engines with hard-coded heuristics.
Shared publiclyView activity