Anthropic Co-founder: Building Claude Code, Lessons From GPT-3 & LLM System Design

In this episode of The Light Cone, Garry Tan interviews Tom Brown, co-founder of Anthropic, about his journey from MIT to co-founding a leading AI company. Brown discusses his early experiences at YC startups, his transition into AI research, and his time at OpenAI, including his work on scaling GPT-3. He shares insights into the founding of Anthropic, the company's focus on AI safety, and the development of Claude, highlighting the importance of internal benchmarks and a model-centric approach. Brown also touches on the challenges of scaling compute infrastructure for AI and offers advice to young people looking to enter the AI field, emphasizing risk-taking and pursuing intrinsically motivating work.