Discovering AI Risks with AIs | Ethan Perez | EAG Bay Area 23
EAG Talks
00:00
Evaluating Data Quality and Addressing Risks in Language Models
This chapter explores the evaluation of data quality in language models and the potential risks and failures associated with them. It discusses techniques such as AI safety via debate or amplification to mitigate these risks and the importance of red teaming to identify failures. The chapter also discusses methods for detecting offensive language, data leakage, and contact information in AI-generated text.
Transcript
Play full episode
