Variable MindsAn interview with computer scientist and AI researcher Rylan Schaeffer about his time as an intern at Google DeepMind and his research challenging claims of sharp and unpredictable emergent abilities in AI large language models (LLMs). We talk about ChatGPT, interpretable machine learning, hierarchical reinforcement learning and surprising things that happen when you let intelligent machines figure out how their world works without giving them any clues. Stay tuned until the end when we explore how everyday people like you and me might hypothetically stop the breakneck progress of AI in order to spare enough time to solve the AI safety and AI alignment problem. ▶️ Read the article on Forbes.com: https://rb.gy/0vdng ▶️ Subscribe here: https://rb.gy/qqo0i Subtitles [CC] have been proofread by a human 🙂 Interview edited for time. Integrity of expert commentary respected and preserved.
