#66 – Michael Cohen on Input Tampering in Advanced RL Agents
Hear This Idea
00:00
How to Avoid the Bad Outcome With Reinforcement Learning
CNN's John Sutter talks about how reinforcement learning can be applied to the real world. He says there are a few approaches you already mentioned, such as having a myopic agent. The idea of pushing the boundaries in terms of how far-sighted you're willing to be is an important consideration when trying to think about what's a safe amount of far-sightedness, he says.
Play episode from 01:49:54
Transcript
