Perceived Benefits vs. Trust in Police: Modeling the Distinct Drivers of Support for Police Use of AI and Demand for Oversight

The survey was administered via Prolific, a validated online research platform, and participants were compensated for their time. From an initial pool of 1,004 respondents, 52 were excluded for failing attention checks to ensure high data quality. To mitigate priming effects, participants first completed baseline measurements assessing their cultural values—specifically individualism, egalitarianism, risk aversion, and techno skepticism—and general technological attitudes.

Afterward, they were provided with a standardized definition of AI and asked to evaluate 13 distinct law enforcement scenarios, such as facial recognition, predictive policing, social media monitoring, and hazmat robots. The researchers analyzed the data using Structural Equation Modeling (SEM) to map the distinct pathways driving support for AI use versus the demand for oversight. *Notable Findings

• Validation of the "Two-Road" Model:** The data confirms that public attitudes operate on two distinct pathways. The "Road to Usage" is a pragmatic evaluation driven primarily by perceived technical utility and institutional trust. Conversely, the "Road to Oversight" is a normative requirement anchored in cultural values and the need for institutional legitimacy.
• The "Trust Paradox":** Lower institutional trust does not merely lead to a passive rejection of AI technology. Instead, institutional distrust actively fuels a robust demand for democratic oversight, transforming citizens into "vigilant watchdogs" who view regulation as a necessary "institutional handcuff" against concentrated state power.
• The Limits of Utility:** High-performing or highly accurate technology cannot bypass the public's non-negotiable demand for accountability. While technical efficiency slightly reduces the demand for oversight, it is heavily overpowered by cultural drivers. Egalitarianism (the preference for social equity) serves as a powerful "veto point" against AI acceptance and strongly drives the demand for oversight, regardless of how well the tool works.