In this episode of Release Notes, the host interviews Connie Fan, product lead, and Danny Tarlow, research lead, for Gemini's coding capabilities. They discuss the journey to creating top-tier coding models, highlighting the shift from competitive programming benchmarks to real-world developer workflows, focusing on data, methodology, and repo context. They emphasize the importance of understanding and addressing developer needs, incorporating feedback from both internal and external sources, and the interconnectedness of coding capabilities with other areas within the Gemini ecosystem. The conversation also touches on the future of coding, the role of AI skeptics, and the potential for AI to solve broader societal problems.
Part 1: Introduction and Goals
Part 2: Integration and Future Vision
Part 3: Addressing Skepticism and Long Context
Part 4: Model Improvement and Personality
Sign in to continue reading, translating and more.
Open full episode in Podwise
