
1 - Adversarial Policies with Adam Gleave
AXRP - the AI X-risk Research Podcast
Is There a Difference Between the Input and the Output of a Robotic Environment?
Am, so moving on a little bit. Ye, seems like a rough challenge. But if i think about these roboic environments, am, the the inputo the policies are like joint, like locations in angles, wrighting up the thread. So to what extent are these like actually importantly different aslike, to what extent ofi actually actually perturbed? Im, like, like uner how? How is some how's what you're doing really that different from an adversarial example? Oh, sure,. yes. O, so the space of possibilities is actually reasonably constrained. We can't have two limbs fur actually intersecting ofa physic similator windn t allow
00:00
Transcript
Play full episode
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.