Research
AI Safety Connect Day, New Delhi, AI Impact Summit 2026
I work on AI ethics, safety, and governance from a philosophical perspective. I am especially interested in how AI systems are evaluated, how they become part of institutional decision-making, and how they affect agency, autonomy, accountability, and legitimacy.
AI evaluation and governance
A growing part of my work focuses on the evaluation of AI systems. I am especially interested in the gap between what evaluations can show under controlled conditions and the claims they are often asked to support in governance, regulation, and deployment decisions.
This includes questions about benchmarks, problem specification, red-teaming, conformity assessment, and safety evaluations. What exactly is being evaluated? What kinds of evidence do different evaluation methods provide? And when is that evidence strong enough to support decisions about deployment, access, monitoring, or regulation?
Agency and Autonomy
A long-standing theme in my work is how AI systems affect human agency and autonomy. I do not treat autonomy as a single value that AI either supports or undermines. Instead, I am interested in the different ways AI systems can shape people’s choices, beliefs, capacities, and sense of control.
Some of this work concerns individual autonomy in interaction with AI systems. Other work looks at agency in more specific settings, including children’s interactions with generative AI, epistemic agency, and cases where AI systems mediate access to information, advice, or institutional decisions.
AI as institutional decision-making
I also work on AI as a form of institutional decision-making. Many AI systems are not best understood as replacing individual human judgment, but as changing how decisions are organised within institutions.
This project examines AI-driven decision-making through concepts usually associated with bureaucracy and administration: rule-boundedness, discretion, consistency, impartiality, accountability, legitimacy, and contestability. I am interested in what this comparison reveals about the promises and limits of AI systems in consequential decision-making contexts.
Earlier work
Earlier work addressed algorithmic profiling and epistemic injustice, community governance in AI, fairness and bias detection, and broader questions in philosophy of science, including scientific objectivity, modelling, and thermodynamics.
Selected Talks and Panel Discussions
Autonomy and Artificial Intelligence, Artificial Intelligence Lab, Brussels, Belgium, 2020
Stakeholder Participation in AI Governance, Bosch Connector, Guadalajara, Mexico, 2020
AI Governance and Ethics, Ethics in AI Seminar Series, University of Oxford, 2020
Beyond near- and long-term: towards a clearer account of research priorities in AI Ethics and Society, AAAI/AIES, New York, 2020
Regulating AI? Challenges and opportunities for the responsible development of AI, EMERTECH conference, EU Delegation in Russia, Moscow, 2019
Expert Panel “Goverance of AI”, Women Leading in AI, London, 2019
Expert, Future Technology Workshop, UK2070 Commission, London, 2019
Expert, Policy Workshop, Centre for Science and Policy, Defence Science and Technology Laboratory, Cambridge, 2019
How anthropocentric is thermodynamics? Sigma Club, London School of Economics, London, 2019
Boltzmann Brains and Simulations - Rethinking the Skeptical Hypothesis, Philosophy of Physics Seminar, Univ. Bonn, 2019
Symposium on Black Holes: Entropy and System Size, British Society for the Philosophy of Science Annual Conference, Oxford 2018
The Role of Information in Black Hole Thermodynamics, Foundational Problems of Black Holes and Gravitation, Munich Centre for Mathematical Philosophy, 2018; European Philosophy of Science Association, Geneva 2019
Resource Theories and Axiomatic Thermodynamics, Philosophy of Physics Conference, University of Western Ontario, 2018
Thermodynamics without Observers?, Conference on the Second Law of Thermodynamics, LMU München, 2017
Black Hole Entropy, how much information do we need?, Sigma Club, London School of Economics, 2018; The Black Hole Initiative Colloquium, Harvard University, 2017
Black Hole Entropy is Entropy (and not Information), Thinking about Space and Time: 100 Years of Applying and Interpreting General Relativity, University of Bern, 2017; 5th International Summer School in Philosophy of Physics, Saig, 2017
A Tale of Two Entropies - defending the von Neumann Entropy, Philosophy of Science Association Biennial Meeting, Atlanta, 2016
Are some quantum interpretations hotter than others?, British Society for the Philosophy of Science Annual Conference, Cardiff, 2016
Other Engagements
Board Member
Oxford AI Society, 2019
Advisor on future governance of AI, UK 2070 Commission, Expert consultation, 2019
Expert Panelist, MentorA.I. Impact Weekend, Oxford Foundry, 8.-10.2.2019