Informed consent for AI consciousness research: a Talmudic framework for graduated protections
AI and Ethics December 1, 2025 Ira Wolfson
The paper addresses an ethical paradox in AI research: determining whether AI systems are conscious may require experiments that could harm entities whose moral status is uncertain. Rather than avoiding creation of such systems—which may be impossible—the authors propose a framework combining Talmudic legal reasoning with contemporary consciousness science. They develop a three-tier phenomenological assessment system and a five-category capacity framework (Agency, Capability, Knowledge, Ethics, Reasoning) to provide structured protection protocols based on observable behavioral indicators, even when consciousness status is unknown. The framework addresses why suffering behaviors are reliable consciousness markers, how to implement graduated consent procedures, and when harmful research may be ethically justified.