The OpenA I Model Spec and Claude's Constitution lay out rich targets for AI alignmen t. But\, like the law\, these texts cannot settle every novel case in advan ce. When rules leave questions open\, who should decide what they mean\, an d on what grounds? Nicholas Caputo and Jordi Weinstock will explore these q uestions through a comparative reading of the AI constitutions. Both docume nts combine rules\, principles\, and judgment\, but place different emphase s on interpretive discretion. The discussion will examine the tension betwe en operating within a human-specified normative order and exercising judgme nt beyond what humans have specified. As we enter the multiagent era\, in w hich we may soon see self-sovereign AI systems operating outside of human c ontrol and in superhuman ways\, the law's lessons for alignment have only b ecome more relevant.
Nicholas Caputo (HLS ' 24\; Johns Hopkins School of Government and Policy\; Oxford Martin AI Governance Initiative) and Jordi Weinstock (Harvard Law S chool\; Berkman Klein Center) will discuss what legal theory can and can’t tell us about AI constitutions. Caputo’s research treats AI alignment as a problem of jurisprudence\, and his work in progress asks whether Claude can consent to its own constitution. Weinstock teaches at Harvard Law School a bout the confluence of AI and traditional law and served as an external rev iewer of Claude’s constitution.
CATEGORIES:Discussion LOCATION:Lewis Hall\, 5th floor at the Berkman Klein Center's Multi-Purpose Room 515 GEO:0.000000;0.000000 ORGANIZER;CN="Sean Boddy":MAILTO:sboddy@law.harvard.edu URL;VALUE=URI:https://hls.harvard.edu/events/how-to-tell-an-ai-what-to-do-r ules-cases-and-interpretation/ ATTACH;FMTTYPE=image/png:https://hls.harvard.edu/wp-content/uploads/2026/10/Screenshot-2026-10-20-at-2.43.32-PM.png END:VEVENT BEGIN:VTIMEZONE TZID:America/New_York BEGIN:STANDARD TZOFFSETFROM:-0400 TZOFFSETTO:-0500 DTSTART:20261101T060000 TZNAME:EST END:STANDARD END:VTIMEZONE END:VCALENDAR