Position: General Intelligence Requires Reward-based Pretraining.

Seungwook Han, Jyothish Pari, Samuel J. Gershman, Pulkit Agrawal

Browse the full ICML paper archive.