I spoke to Trevor Chow about existential risks from AI and techniques to align artificial intelligence with human goals. Specifically we talked about
- An introduction to existential risk from Artificial Intelligence
- Existing methods for alignment of AI models
- Why RLHF might fail in large language models
- Whether interpretability research might scale?
- New methods being developed to make larger models safer
- Regulatory frameworks for the future of AI
--- Send in a voice message: https://podcasters.spotify.com/pod/show/pradyumna-sp/message