Ari Deller
PhD Candidate in Philosophy · University of Cambridge
I work in epistemology, ethics, and the philosophy of artificial intelligence, and on AI safety and governance.
I am a PhD candidate in Philosophy at the University of Cambridge, where I am a Peterhouse Graduate Scholar supervised by Jessie Munton and Richard Holton. In Spring 2025 I was a Visiting PhD Researcher at MIT. Before Cambridge I read for the BPhil at Oxford and an MA at Edinburgh.
My research concerns epistemic agency and how it comes under two practical pressures:
- the demands of ethics;
- the proliferation of increasingly capable artificial intelligence.
My doctoral thesis, Epistemic Agency and Practical Pressures, pursues this theme across the grounds of epistemic normativity, the morality of belief, and the epistemic stakes of living alongside persuasive AI.
Current focus
AI safety & governance
Alongside my doctoral work, I hold fellowships in AI governance and safety, currently as an ERA:AI Governance Fellow and a Fellow of the Cambridge AI Safety Hub. I work on AI persuasion and harmful manipulation:
- defining and assessing harmful manipulation for the EU AI Office and frontier-lab safety frameworks;
- developing HumanRationalityBench, a benchmark tracking whether large language models degrade or improve human rationality, funded by a Cosmos Institute grant.
A paper, a policy memo, and a draft of the benchmark are available on request.
Fig. 1 — Click a hatched overlap to reveal the work at that intersection.
Research interests
- Epistemology — epistemic normativity, epistemic agency, and defeat
- The ethics of belief — doxastic wronging and the morality of belief
- Philosophy of artificial intelligence
- AI safety and governance — AI persuasion, harmful manipulation, and human rationality
- Connections to the philosophy of mind and the self — care, authenticity
Recent work
The Epistemic Costs of Super-Persuasive AI
Argues that AI which is highly persuasive but not reliably truthful imposes real epistemic costs — chief among them, a pervasive undercutting defeat of beliefs we form through argument. Read more →