AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
1min snip
Dario Amodei (Anthropic CEO) - Scaling, Alignment, & AI Progress
Dwarkesh Podcast
Uncovering Mechanistic Interpretability in Machine Learning
There is a mystery around how machine learning models like neural networks, once unable to perform addition, suddenly start grasping it. Even with improved accuracy, the mechanism behind this leap is not clear. The phenomenon of models becoming better at tasks before reaching the right answer hints at a continuous underlying process. The concept of a preexisting circuit that gains strength or a weak circuit that improves is puzzling researchers. The quest for mechanistic interpretability seeks to unravel whether certain abilities of models will remain hidden even with increased scale.
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