Open materials supporting this research, by stream:
My research sits at the intersection of philosophy, artificial intelligence, and governance. It spans four connected streams — from the epistemics of frontier models, to the formal foundations of fairness and trust, the institutional design of AI accountability, and the ethics of data and AI in health.
How large language models form and revise beliefs, attribute sources, and express (mis)calibrated confidence — and how to evaluate and constrain these behaviours. Topics: LLM belief-formation, source-attribution bias, coherence bias, calibration, epistemic constitutions, and model self-correction. This stream connects classical epistemology to the evaluation of frontier AI systems.
The logico-mathematical structure of trust, fairness, and explanation in algorithmic systems. Topics: formal models of trust, counterfactual fairness, equality of opportunity, calibration as a fairness requirement, explainability, and the robustness of explanations over time. Representative work: “How Much Do You Trust Me?” (Synthese 2023); “In AI We Trust Incrementally” (Philosophy & Technology 2020); “A Moral Framework for Understanding Fair ML” (ACM FAT* 2019); “Fair Equality of Chances for Prediction-Based Decisions” (Economics & Philosophy 2023); “Is Calibration a Fairness Requirement?” (ACM FAccT 2022).
Institutional design for the responsible deployment of AI in the public and private sector. Topics: public-sector AI, algorithmic audits, institutional accountability, democratic oversight, systemic-risk regulation (Digital Services Act and EU AI Act), and deployment governance. Representative work: “Towards Accountability in the Use of AI for Public Administrations” (AAAI/ACM AIES 2021); “Regulating the Undefined: Addressing Systemic Risks in the Digital Services Act” (Philosophy & Technology 2025); the AlgorithmWatch Impact Assessment Tool for public-sector automated decision-making.
The ethics of data and predictive systems in medicine and public health. Topics: the digital phenotype, data cooperatives and data ownership, clinical prediction, fairness in health, and digital contact tracing. Representative work: “The Digital Phenotype” (Philosophy & Technology 2019); “Towards Rawlsian ‘Property-Owning Democracy’ Through Personal Data Platform Cooperatives” (CRISPP 2023); “How to Fairly Incentivise Digital Contact Tracing” (Journal of Medical Ethics 2021); the “Fair Predictions in Health” project (Marie Skłodowska-Curie, Politecnico di Milano).
