aceModerate evidence
More
In a 648-person experiment, the same police burglary report felt less fair from a chatbot
Study title: Reporting crime to police via human and AI live chats: the salience of procedural justice and outcome favourability
- Evidence type
- Randomized controlled trial
- Sample
- n = 648
- Sources
- 1 sourcePeer-reviewed
- Published
- 2 min read
- Truvace record
- TRV-2026-2495cite or check it
Study title: Reporting crime to police via human and AI live chats: the salience of procedural justice and outcome favourability
An online experiment with 648 participants found an AI penalty in how people judge police contact. Participants read a simulated exchange in which a burglary was reported by live chat to either a human officer or an AI chatbot. The same behaviour was rated less fair and less satisfactory when the chatbot delivered it.
Read the study abstract
As police services adopt automated systems for crime reporting, questions arise about how these technologies affect public perceptions of fairness and legitimacy. This study examines how people evaluate police–public interactions conducted via human or AI-operated live chat. In an online experiment (N = 648), participants read a simulated exchange in which a burglary was reported to either a human officer or an AI chatbot. We manipulated operator type (human vs. AI), procedural justice (fair vs. unfair), and out…
The problem
In an online experiment of 648 participants, identical police behaviour in a simulated burglary report was rated less fair and less satisfactory when delivered by an AI chatbot than by a human officer.
In brief
The result matters as police services adopt automated systems for crime reporting. Perceived fairness and satisfaction are tied to public views of police legitimacy, and the study suggests that handing contact to a chatbot can lower both even when the treatment is identical.
Fair treatment still counted. Procedural justice shaped fairness and satisfaction for both human and AI operators. The experiment used simulated exchanges, so it measures reactions to scenarios rather than real reports.
Main points
- The online experiment had 648 participants who read a simulated burglary report made through live chat.
- Reports went to either a human officer or an AI chatbot.
- Researchers varied operator type (human or AI), procedural justice (fair or unfair) and outcome (active or passive).
- Participants rated perceived fairness, satisfaction and willingness to use the reporting method.
- Procedural justice shaped fairness and satisfaction in both the human and AI conditions.
- Identical behaviour was rated less fair and less satisfactory when a chatbot delivered it.
- Negative effects of a passive outcome were reduced when the treatment was fair.
The full story
The study, published in Police Practice and Research, tested how people evaluate police–public interactions by live chat. Participants read a simulated exchange in which a burglary was reported to either a human officer or an AI chatbot, and the researchers varied three factors: operator type, procedural justice and outcome.
Participants rated perceived fairness, satisfaction and willingness to use the reporting method. Fair treatment raised fairness and satisfaction ratings in both conditions, but the chatbot was rated lower than the human officer for identical behaviour.
Outcome also influenced responses, comparing active and passive police outcomes. Its negative effects were mitigated by fair treatment. The authors say the findings point to the importance of procedural justice and its attenuation in AI-mediated policing.
Sources
1 on file- Police Practice and Researchdoi:10.1080/15614263.2026.2741740
Talk it over
Who carries this cost, and what would have to change to prevent it?
In a 648-person experiment, the same police burglary report felt less fair from a chatbothttps://truvace.com/item/reporting-crime-to-police-via-human-and-ai-live-chats-the-salience-of-procedural
Who carries this cost, and what would have to change to prevent it?
The problem In an online experiment of 648 participants, identical police behaviour in a simulated burglary report was rated less fair and less satisfactory when delivered by an AI chatbot than by a human officer.
Answer in the threadHow do we know? What can this kind of evidence show, and what can’t it?
The evidence This rests on a randomized controlled trial with a sample of 648, from Police Practice and Research.
Answer in the threadWhat does it leave open? What else would you want to know before acting on it?
The record’s own caveat The study used simulated written exchanges in an online experiment, so participants evaluated a scenario they read rather than a burglary they reported to police.
Answer in the threadHave you seen AI used this way where you study, work or live? Did it go the same way?
Answer in the thread
The questions are the same on every record. The details under them are taken from this record as written.
What readers say