Report
A policy reportedly transferred from GTA to horseback movement in Red Dead Redemption without retraining
A State of AI editor’s cut describes Odyssey’s exploration of world prediction and action across driving, robotics and games.
TLDR
Sharing an excerpt from a State of AI editor’s cut, its author says Odyssey is exploring whether learning to predict the world can also help a model act in it. They highlight one selected example in which a policy transferred from GTA to horseback movement in Red Dead Redemption without retraining.
Combined views
641
1 Source, first seen ago
