AI Safety Audits
Jeremie discusses the importance of external audits in AI safety, highlighting the work done by the alignment research center and the potential risks of AI systems seeking power. He emphasizes the need for effective strategies, like constitutional AI, to ensure that AI aligns with safe human goals despite the inherent limitations of human feedback mechanisms. The conversation touches on the challenges of both inner and outer alignment in advanced AI systems.In this clip
From this podcast
Super Data Science: ML & AI Podcast with Jon Krohn
668: GPT-4: Apocalyptic stepping stone? — with Jeremie Harris
Related Questions
Can AI motivations be shaped as discussed in the episode Jeremie Harris: Realistic Alignment and AI Policy and the clip The Challenge of AI Objectives?
Should we be concerned about AI risks as discussed in the episode 888: Marc Andreessen | Exploring the Power, Peril, and Potential of AI and the clip AI Alignment Concerns, as well as in the episode Jeremie Harris: Realistic Alignment and AI Policy and the clip Shifting Public Perception?
Should we be concerned about AI risks as discussed in the episode #176 - SearchGPT, Gemini 1.5 Flash, Lamma 3.1 405B, Mistral Large 2 and the clip OpenAI's Safety Commitments, as well as in the episode 888: Marc Andreessen | Exploring the Power, Peril, and Potential of AI and the clip AI Alignment Concerns, and the episode Jeremie Harris: Realistic Alignment and AI Policy and the clip Shifting Public Perception?