AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
1min snip
The Alignment Problem From a Deep Learning Perspective
AI Safety Fundamentals: Alignment
Impact of Reward Misspecification and Fixation on Feedback Mechanisms on Goal Alignment
Consistent reward misspecification can reinforce misaligned goals and lead to policies learning to maximize reward rather than fulfilling aligned goals. Intrinsic curiosity reward functions may lead policies to consistently pursue the goal of discovering novel states, conflicting with aligned goals. Additionally, goals can be correlated with rewards due to fixation on feedback mechanisms, rather than the content of the reward function.
00:00
Transcript
Episode notes
Get the Snipd
podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any
moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share
& Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode