Adam Davies

About

Hi there! 👋 I’m Adam, an AI researcher at Microsoft’s IDEAS Research team. I recently (in May 2026) graduated from UIUC (opens in new tab) with my PhD in CS (opens in new tab).

My research vision is to help build the foundations of a cognitive science of AI (opens in new tab), studying how foundation models like LLMs learn, represent, and process information. My work (opens in new tab) leverages insights from cognitive science to develop new approaches to mechanistic interpretability that can help us better understand machine cognition at scale.

My current research agenda is centered around bidirectional theory of mind (opens in new tab) between AI agents and human users – that is:

• How do AI systems internally represent and reason over the mental states of human users? For instance: when, how, why do they (mis)understand our goals, intentions, or preferences? How can we proactively surface and correct such misunderstandings to promote more productive and safe interactions with AI?
• How can we, as human researchers (or everyday users), better understand why AI systems behave the way they do? What do they (mis)understand about us, our values (individually and societally), and the world? When can(‘t) we trust them to act autonomously on our behalf?