UW's AI Vice Provost on Open Models & Future of Work

Summary

University of Washington's Vice Provost for AI, Noah Smith, reveals how faculty are creatively integrating AI into classrooms, from interactive logic textbooks to personalized chatbot experiences. Smith's team at the Allen Institute for AI champions fully open language models like Olmo, sharing everything from code to data, which is crucial for researchers wanting to study training processes or fine-tune models with their own data. This contrasts with proprietary models, offering greater control and flexibility, especially for regulated industries. He emphasizes that AI literacy and critical thinking, core university values, are key to preparing students for a future where agents handle routine tasks, ensuring human agency remains paramount. Smith also discusses the university's proactive approach to AI governance, with committees working to establish principles and structures, highlighting how AI is impacting all aspects of campus life, not just research and teaching. The full conversation is worth your time for its insightful exploration of AI's educational impact, the benefits of open-source development, and a hopeful vision for AI enhancing human control over work and life.