AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
4min chapter
168 - How to Solve AI Alignment with Paul Christiano
Bankless
The Importance of Learning to Be Honest
There are lots of cases in which it is incentivized to lie or mislead the human. There's a gap between like lying that will get caught and lying that won't get caught And so you can ask if we train neural nets do they exhibit this kind of switch abruptly? If they get put in a position where they could get away with something really sinister Will they then do it? I think one reason for optimism right now is no one has ever really exhibited that phenomenon in a convincing way. It just is much easier as your models get more competent Like it's only recently that we've trained models which are actually able to understand the mechanics of their training process.
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