"Scientific Wellness and the Question of Who Knows You Best": Self-tracking technologies increasingly promise to help us understand ourselves better, using data from wearables, apps, and other digital systems to guide decisions about health, behavior, and well-being. But these developments raise a deeper question: who knows you best — you, your doctor, or your data? In this talk, I will discuss the idea of scientific wellness as an emerging framework for combining personal data, experimentation, and intelligent feedback to support self-understanding and behavior change. I will introduce N-of-1 trials as one rigorous approach to personalization, and argue that bringing such methods into everyday life could reshape not only how we track ourselves, but also how we think about expertise, agency, and autonomy in digital health. — Dominik Meier

"Training Natively Interactive Models": Large language models are interactive artifacts, but interaction is baked into them almost offhandedly, both for deployment and learning, and this has far-reaching consequences: rather than supporting the naturalness of human interactions, interactions with LLMs are stilted turn-taking experiences, a far cry from the dynamic full-duplex (i.e., concurrent input/output) nature of human interaction. While reinforcement learning methods utilize interactions, they remain mostly applicable to narrow domains due to their dependence on simplified rewards or the risks of reward hacking. I will spend most of the talk describing full-duplex speech LLMs and the methods I designed to post-train them, showing consistent improvements in conversational engagement, safety, and factuality; and in the second part, I will discuss an RL benchmark for visuo-linguistic reasoning I created, focused on designing robust reward signals that are not susceptible to reward hacking or require significant task simplifications. — Anne Wu