Discovering Reward Functions
Marcus explains how in simple tasks, the reward function can be easily defined, but in more complex problems, it becomes challenging. He shares examples of elevator control and building general agents. The human should give the reboot on the fly in the case of general agents.In this clip
From this podcast

Lex Fridman Podcast
Marcus Hutter: Universal Artificial Intelligence, AIXI, and AGI | Lex Fridman Podcast #75
Related Questions