"Till I can get my satisfaction": Open Questions in the Public Desire to Punish AI

The public blames AI itself for harms, not just developers: Research shows people are willing to ascribe blame directly to AI systems regardless of whether this seems "illogical" from a technical standpoint, suggesting public attitudes toward AI responsibility differ fundamentally from how responsibility is currently assigned in law and policy.

Unmet public expectations for punishment may leave victims unsatisfied: The paper warns of a potential "satisfaction gap" where legal consequences (whether AI systems are punished or not) fail to match what the public expects, potentially leaving victims of AI harm with lingering feelings of injustice that could undermine public trust in AI governance.

We don't yet understand what "punishing AI" would even mean: The paper identifies a critical knowledge gap—there is virtually no research on what forms of punishment the public actually expects for AI systems, making it impossible to design accountability mechanisms that would satisfy public desires for justice.

This is an interdisciplinary problem requiring urgent research: Understanding public expectations about AI punishment requires insights from psychology, human-computer interaction, philosophy, ethics, and law working together—no single field has adequate answers yet.

Public satisfaction matters for responsible AI governance: How societies handle accountability for AI harm isn't just a technical or legal question; it's fundamentally about whether the public feels their sense of justice has been addressed, which affects broader acceptance and trust in AI systems.