The Origins Podcast with Lawrence Krauss cover image

Scott Aaronson: From Quantum Computing to AI Safety

The Origins Podcast with Lawrence Krauss

00:00

Controlling Language Models and Risks of False Positives

This chapter explores the risks of false positives in discriminator models and different approaches to controlling language models, including statistical watermarking and temperature adjustment. The authors discuss the limitations of certain approaches and suggest the need for watermarking at the semantic level.

Transcript
Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app