"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis cover image

Dodging Latent Space Detectors: Obfuscated Activation Attacks with Luke, Erik, and Scott.

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis

NOTE

Exploring Harmfulness Monitoring in AI Models

This chapter examines three case studies focused on harmfulness monitoring in AI models, highlighting the use of supervised probes to identify harmful queries and responses. It discusses input space perturbation tactics and the technical intricacies of probes, such as multi-layer perceptrons, to enhance the understanding of their role in detecting harmful outputs.

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