We're sunsetting PodQuest on 2025-07-28. Thank you for your support!
Export Podcast Subscriptions
back
“Ctrl-Z: Controlling AI Agents via Resampling” by abhatt349, Buck, Adam Kaufman, Cody Rushing, Tyler Tracy
24:21
Share
2025/4/16
LessWrong (30+ Karma)
AI Chapters
Transcribe
Chapters
What is the Multi-Step Control Problem?
How Does Resampling Work?
Selection Pressure: Can It Force Consistent Attacks?
Value of Information: How Does It Improve Attack Detection?
What Are the Important Limitations of These Techniques?
Summary of Results: What Did They Find?
Core Takeaways: What Are the Key Insights?
Shownotes
Transcript
No transcript made for this episode yet, you may request it for free.