34 - AI Evaluations with Beth Barnes
AXRP - the AI X-risk Research Podcast
Evaluating AI Models: Risks and Realities
This chapter explores the complexities of capabilities evaluations of AI models, focusing on issues like sandbagging and misrepresentation during training. The conversation covers the challenges in understanding AI behavior, the significance of establishing a capability scale, and the importance of addressing risks associated with both lower and higher-tier models. Additionally, it highlights the need for robust evaluation methods and buffer systems to ensure safe and effective AI development.
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.
