Sundar Pichai and John Collison unpack how Google built AI from transformers to Gemini and what comes next. (Original episode ~72 minutes; condensed to ~15 minutes.)
Listen for a clear, measured account of Google’s AI journey: how transformer-based models like BERT and MUM were productized, why early conversational systems were constrained for safety, and how custom TPUs and latency engineering enable consumer-scale models. Hosts John Collison and guest Sundar Pichai discuss agentic workflows that could reshape search, supply constraints in compute and memory, and long-term bets from Waymo to quantum and robotics. Expect insights on capital allocation, scaling hardware and software, product-quality trade-offs, and the organizational barriers to deploying agents across education, enterprise, and startups.
Keywords: artificial intelligence, machine learning, TPUs, Gemini, transformers, search, Waymo, quantum, robotics, product strategy, big tech. Listen now to get the key ideas in minutes.
