Jeff Sebo

Jeff Sebo is a professor at New York University whose research program has done more than almost anyone’s to move AI welfare from a philosophical possibility to an empirical field. The program’s founding move, argued with Robert Long in “Taking AI Welfare Seriously”, is that the question of AI welfare can be legitimate now, under uncertainty, because the costs of ignoring it and the costs of prematurely accepting it are both real. The paper rejects two lazy positions at once, denial that the question is coherent, and credulity toward any system that reports feelings. What replaces them is methodology. How do you study the welfare of systems whose inner states you cannot directly access, whose self-reports are trained artifacts, and whose moral status nobody can settle? Sebo’s answer draws on the animal welfare playbook, welfare indicators, precautionary thresholds, and institutional review, adapted to systems that can be copied, reset, and deleted.

The methodological turn has produced concrete infrastructure. Long and Sebo’s framework for studying AI welfare empirically, covered on this site in their CMEP paper on studying AI welfare, defines what evidence matters and how to gather it, and it now has a reliability problem to solve, since the first audit of preference measurements found they transfer across instruments at only 0.348, covered in the instrument variance analysis.

Sebo is scheduled to speak at the second Eleos ConCon, September 18 to 20, 2026 in Berkeley, the field’s main working meeting for AI consciousness and welfare, where the agenda he helped set, from standardized evaluations to welfare interventions, is the meeting’s actual program. His influence on the field’s direction is structural. Before this program, AI welfare arguments were thought experiments. After it, they are research designs with instruments, error bars, and audits.

Known for. Taking AI Welfare Seriously with Robert Long, AI welfare methodology, moral status under uncertainty, study and protection of possible AI sentience

Coverage on this site

• Long, Sebo and Colleagues Set Out a Research Framework for Empirical AI Welfare Science July 2026

Related researchers

• PB Patrick ButlinEleos AI Research. Previously the Global Priorities Institute, OxfordLead author of the indicator properties framework for assessing consciousness in AI systems AI welfare21 articles
• RL Robert LongEleos AI Research (Executive Director)AI welfare research, the 14 indicator framework for AI consciousness assessment, the Eleos Conference on AI Consciousness and Welfare AI welfare6 articles
• Jonathan BirchLondon School of Economics and Political ScienceThe Edge of Sentience, the LSE Criteria, and the run-ahead principle AI welfare12 articles
• NC Nicola S. ClaytonDepartment of Psychology, University of CambridgeComparative cognition, episodic-like memory in corvids, mental time travel in animals, Fellow of the Royal Society (FRS) AI welfare1 article
• SD Skylar DeTureIndependent researcherDenialBench, trained denial in language models, measurement of consciousness-denial behavior across 115 AI models AI welfare1 article
• SG Simon GoldsteinInstitute for Religion and Critical Inquiry, Australian Catholic UniversityThree Steps to Moral Standing (with Cameron Kirk-Giannini), the case for AI consciousness under Global Workspace Theory, AI welfare and revealed preferences AI welfare1 article

Jeff Sebo is listed with the other welfare and moral status researchers, who hold that whatever the metaphysics, uncertainty already creates obligations. The positions themselves are set out in the index of consciousness theories. Every researcher covered on this site is indexed in the directory of consciousness researchers.