Communications of the ACM
Did DeepMind Just Take a Big Step Toward More Human-Like AI?
This is very different from software DeepMind has created previously in which an algorithm could be trained to perform at superhuman levels on a variety of games. In those cases, the algorithm had to be retrained for each game.
DeepMind researchers have created software agents that navigate a three-dimensional game world called XLand, where the agents learn from experience, by trial and error, how to play a vast number of simple games. Based on that experience, they were able to apply those learned skills and strategies to new games, without specialized training. The bots had acquired general skills useful for playing games in that simulated world, even ones they had not encountered before.
While AGI may still be years or decades away, or may never arrive, the possibility suddenly seems closer than it did before.
From Fortune View on AI
View Full Article – May Require Paid Registration
No entries found