Philosopher-Technologist
Academic Technology Consultant, Williams College
Selected Work
-
2025
Humanity's Last Exam
Co-author on a multi-modal LLM benchmark published in Nature (2026). Project of The Center for AI Safety.
-
2025
Persuadable Machines
Public talk on using classical rhetorical techniques to bypass safety guardrails in generative AI, demonstrating the intersection of ancient philosophy and modern AI security.
-
2025
The Inconsistency Critique
Academic article examining epistemic practices and AI testimony about inner states, arguing that our selective treatment of AI output as testimony demonstrates structures of prejudgment.
-
2024
NIST-Sponsored Red-Teaming
Invited participant in CAMLIS AI red-teaming exercise sponsored by NIST and Humane Intelligence. Tested generative AI security using the NIST AI 600-1 framework.
-
2025
The Testimony Problem
An in-progress academic monograph examining epistemic challenges in AI moral status, developed through sustained dialogue with Claude as an interlocutor.
-
2025
Financial Risks of GenAI
Invited talk at Institute of Management Accountants on financial risks of generative AI in business contexts, translating technical risks for financial professionals.