Ongoing Projects
Our research spans those who build and procure systems and those who use and are affected by them. Our projects interrogate the infrastructural and social conditions that make AI systems possible and how they are sustained and governed in practice, with particular attention to the judgments, tensions, and adaptations that shape ongoing implementation.
Conversational AI for Mental and Emotional Care
We study how people turn to general-purpose AI chatbots for emotional support, companionship, and mental health care, often in the absence of affordable or accessible professional alternatives. Our research explores how these relationships develop over time, following the reliance they generate, the harms that accumulate, and where existing oversight frameworks fall short.
AI, Science, and the Future of Knowing
Through close observation of laboratory practice, we trace how AI tools are reshaping the habits of scientific reasoning: what counts as evidence, how expertise is exercised, how judgment is taught, and where claims to knowledge become credible or begin to unravel. We translate these findings into practical interventions for funders and research institutions, opening new ways to evaluate and govern scientific work as AI becomes part of its everyday practice.
Benchmarking and Evaluating AI
This project examines how AI systems come to be measured and compared in ways that make them appear trustworthy. Our research follows evaluation practices and the making of benchmarks, tracing how complex domains are translated into datasets, tasks, prompts, metrics, and standards. Across this work, we ask how evaluation can remain accountable to the forms of expertise and situated contexts that make benchmark results meaningful.
