I am developing rigorous, analytic frameworks for measuring and comparing information and belief states—work that travels across agent–agent comparison, text–text comparison, and the evaluation of AI outputs.
My dissertation research centers on formal epistemology and philosophy of AI, working with David Thorstad to create methodologies that bridge technical implementation with normative reasoning. My background in classical Islamic philosophy, particularly textual source analysis, brought me into AI as a tool for quantifying similarity and uncovering overlooked texts—work that directly informs my current focus on similarity quantification and semantic distance in formal epistemology.
Epistemic Distance & Belief Comparison
Developing hierarchy-sensitive methods for comparing non-ideal agents' belief systems using information-theoretic tools including Bregman-style divergences, extending accuracy-first approaches to more practical settings.
Credence-Preservation in AI Evaluation
Creating evaluation frameworks for automatic text summaries that assess the beliefs they induce rather than surface overlap alone, with applications to high-stakes policy contexts.