AXRP - the AI X-risk Research Podcast cover image

35 - Peter Hase on LLM Beliefs and Easy-to-Hard Generalization

AXRP - the AI X-risk Research Podcast

NOTE

Understand Beliefs Through Task Clarity

Detecting beliefs in language models requires careful consideration of assumptions regarding understanding, truthfulness, and honesty. Clarifying the task at hand is crucial, especially in terms of how truth is framed within contemporary scientific perspectives. A study compared basic factual prompts, easily answerable by young children, to more complex questions to highlight differences in answering capabilities. This approach seeks to differentiate between domain knowledge, mathematical ability, and the presentation style of questions to better understand how models interpret and respond to inquiries.

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.
App store bannerPlay store banner