2 ms·
We're very, very far from a good discovery AI. Currently world is at a stage of "not get trapped in a local minimum when optimizing known decisions", which is v
by AstralStorm 9y ago
We're very, very far from a good discovery AI. Currently world is at a stage of "not get trapped in a local minimum when optimizing known decisions", which is very far from actually generating decisions.
Latest research in sparse reward games is a good step forward, as research is generally one.