Talk:Automated decision-making

[Untitled]

This page is a work in progress but is hopefully beyond stub status now but there is lots more to add. Please note although some consider ADM to be the same as, or a subset of, AI this is not the case - AIs may do things besides provide decision-making and decision making systems may not use AI. ADM is a major area of social and political life that warrants its own page and should not be redirected to AI.

Disclaimer: I'm working on this page as part of my role as Wikimedia in Residence at the ARC Centre of Excellence for Automated Decision-Making and society. The aim is to develop more quality content about ADM and related technical and social issues. See the project page for more info Amanda Lawrence 03:58, 6 October 2021 (UTC)

Possible addition: can systems document human review?

I'm the author of a source that might be relevant here, so flagging on the talk page rather than editing directly (WP:COI). The article cites GDPR Article 22 and the UK's Data (Use and Access) Act 2025 requirement for "meaningful human review," but doesn't address whether deployed systems can actually produce a record showing that review occurred. A dataset reading ten widely deployed AI agent frameworks against twenty record-keeping requirements found that most cannot record who reviewed a decision, deposited with a DOI at https://machinetestimony.org/census/2026-09/. Would a sentence on this fit the Explainability or Legal issues section, or is this out of scope for the article? Machine Testimony (talk) 17:15, 15 September 2026 (UTC)