Logo image
Back
What can the Post Office ‘Horizon’ scandal teach about Artificial Intelligence deployment?
Conference paper   Peer reviewed

What can the Post Office ‘Horizon’ scandal teach about Artificial Intelligence deployment?

Chris Baber, Brandon King, Paul Salmon and Yihao Jiang
Contemporary Ergonomics and Human Factors 2025, pp.224-232
Ergonomics & Human Factors Conference, 2025 (Burton upon Trent, United Kingdom, 28-Apr-2025–30-Apr-2025)
Chartered Institute of Ergonomics and Human Factors (CIEHF)
2025
url
https://publications.ergonomics.org.uk/uploads/What-can-the-Post-Office-%E2%80%98Horizon-scandal-teach-about-Artificial-Intelligence-deployment.pdfView
Published Version Open

Abstract

Post Office Horizon Accimaps Socio-Technical Systems Transparency
In this paper we use the Post Office Horizon scandal. Broadly, accounting errors arising from a computer system were falsely attributed to Subpostmasters who were subsequently accused of theft, fraud, and false accounting. Over 700 prosecutions have been successfully appealed and convictions quashed. Coupled with severe technical shortcomings, there was confirmation bias in the decisions of Post Office Limited to prosecute Subpostmasters and an assumption in UK Law that computers do not make mistakes (unless proven otherwise). From a review of evidence and newspaper reports of specific cases in the Post Office Horizon scandal, we construct and analyse Accimaps. We argue that a common problem across these cases is how different sorts of ‘black box’ of the Horizon system meant that it lacked transparency for all stakeholders.

Details

Metrics

51 Record Views
Logo image