AI Safety Fundamentals: Alignment

83 Episoder

1. Public by Default: How We Manage Information Visibility at Get on Board

   Publisert: 12.5.2024
2. Writing, Briefly

   Publisert: 12.5.2024
3. Being the (Pareto) Best in the World

   Publisert: 4.5.2024
4. How to Succeed as an Early-Stage Researcher: The “Lean Startup” Approach

   Publisert: 23.4.2024
5. Become a Person who Actually Does Things

   Publisert: 17.4.2024
6. Planning a High-Impact Career: A Summary of Everything You Need to Know in 7 Points

   Publisert: 16.4.2024
7. Working in AI Alignment

   Publisert: 14.4.2024
8. Computing Power and the Governance of AI

   Publisert: 7.4.2024
9. Emerging Processes for Frontier AI Safety

   Publisert: 7.4.2024
10. Challenges in Evaluating AI Systems

    Publisert: 7.4.2024
11. AI Control: Improving Safety Despite Intentional Subversion

    Publisert: 7.4.2024
12. AI Watermarking Won’t Curb Disinformation

    Publisert: 7.4.2024
13. Interpretability in the Wild: A Circuit for Indirect Object Identification in GPT-2 Small

    Publisert: 1.4.2024
14. Towards Monosemanticity: Decomposing Language Models With Dictionary Learning

    Publisert: 31.3.2024
15. Zoom In: An Introduction to Circuits

    Publisert: 31.3.2024
16. Weak-To-Strong Generalization: Eliciting Strong Capabilities With Weak Supervision

    Publisert: 26.3.2024
17. Can We Scale Human Feedback for Complex AI Tasks?

    Publisert: 26.3.2024
18. Machine Learning for Humans: Supervised Learning

    Publisert: 13.5.2023
19. Four Background Claims

    Publisert: 13.5.2023
20. Biological Anchors: A Trick That Might Or Might Not Work

    Publisert: 13.5.2023