Model Evaluation For Extreme Risks of AI | Google DeepMind and OpenAI Paper
🔥 Get my A.I. + Business Newsletter (free): https://natural20.com/ #ai #google #deepmind So today, DeepMind drops a new paper, an early warning system for novel AI risk. New research proposes a framework for evaluating general purpose models against novel threats. There's some pretty big implications here. RELATED VIDEOS: "We have no moat" - Google https://www.youtube.com/watch?v=h5C7pxQ8wY0 Tree of Thoughts https://youtu.be/BrjAt-wvEXI Governance of Superintelligence by OpenAI https://youtu.be/EX6UuNIpI04 LINKS: Jack Clark Twitter: https://twitter.com/jackclarksf GPT-4 System Card https://cdn.openai.com/papers/gpt-4-system-card.pdf Tree of Thoughts: https://arxiv.org/pdf/2305.10601.pdf Governance of superintelligence https://openai.com/blog/governance-of-superintelligence OpenAI Hide and Seek https://openai.com/research/emergent-tool-use Live View of 25 AI Agents in a Village: https://reverie.herokuapp.com/arXiv_Demo/ Less Wrong Post: https://www.lesswrong.com/posts/FdQzArWhERh4YZqY9/deepmind-model-evaluation-for-extreme-risks DeepMind Blog: https://www.deepmind.com/blog/an-early-warning-system-for-novel-ai-risks Model evaluation for extreme risks https://arxiv.org/abs/2305.15324 TIMELINE: [00:00] Intro [00:31] Recap [01:27] OpenAI Proposal [02:21] Google DeepMind Paper [04:46] Abrupt Emergence [07:48] Existing Benchmarks [09:08] "Frontier" [13:35] Alignment Eval [15:09] Dangerous Capabilities [19:12] Safety Workflow [21:25] Power Seeking AI [24:05] Unknown Capabilities [30:43] Hazards
