In this episode, Diyi Yang discusses how large language models can meaningfully augment human capabilities in both work and well-being. She explains her research on building AI partners and mentors that help people practice supportive communication skills through realistic, coached interactions—an approach designed not to replace human connection but to strengthen it. She also describes her team’s landmark “Future of Work with AI Agents” study, which mapped how workers across many occupations want AI to automate or augment their tasks, revealing major gaps between what workers want, what AI can do, and where research is currently focused. A central idea throughout is “human agency”: understanding how humans and AI share control across tasks and why the right granularity—looking at workflows, not whole jobs—is essential for successful augmentation.

The conversation also explores emergence, the creation of new roles and opportunities when humans and AI collaborate rather than compete. Yang shares insights from studies on AI-generated research ideas, revealing an “ideation-execution gap” where AI proposals appear highly novel but often fall short in practical implementation. She emphasizes the importance of understanding hidden human skills through workflow auditing, designing AI that enhances long-term well-being rather than short-term engagement, and building systems that surface latent human capabilities. Ultimately, she is optimistic about a future where humans and AI expand what is possible together—and where conversations, creativity, and domain expertise remain at the center of meaningful progress.