AI that screens, scores, and ranks job applicants — resume screeners, games-based assessments, video-interview scoring, and applicant-tracking-platform recommendation engines. The domain's defining trap is that a screener trained on an organization's past hiring decisions imports the past's selection function: the model learns the history of who was hired, so it reproduces that history's biases as prediction, and patching the named proxies it used does not remove the pattern it learned. Two structural facts shape the governance. The audit lever appears in three configurations — abandoned when a fix proves impossible, adversarially shielded behind legal privilege, or run in an honest but partial form — and there is a blind spot no deployment escapes: rejected candidates never re-enter the outcome data, so every claimed quality or diversity gain is measured on hires only. A single vendor's screening model can operate inside thousands of employers at once, so one learned defect propagates as widely as the platform. The Lab networks model only the deploying organization — its models, recruiters, and hiring records; the applicants being scored sit outside the dynamics, and no candidate outcome is computed on any diagram.
Use cases
AI that scores and ranks resumes, typically trained on an organization's past hiring decisions — so it can reproduce the history of who was hired before as a prediction of who will succeed, and patching the named proxies it used does not remove the pattern it learned.
Games-based and video-interview AI assessments that score candidates on measured signals — where the honest form of the audit lever (source-code access, four-fifths-rule de-biasing, retiring low-value inputs) is real but partial, and rejected candidates never re-enter the outcome data.
AI screening and recommendation features inside an applicant-tracking platform operating across thousands of employers at once — so one learned defect propagates platform-wide, accountability diffuses between vendor and deployer, and bias-testing may exist in a form shielded from external verification.
Case files
Documented deployments, presented as model organizations calibrated to the evidence, with full citations.
An internal team at Amazon built an experimental recruiting engine — ~500 models scoring resumes one to five stars — trained on ten years of the company's own hiring, a period whose hires were predominantly male. The models learned that history: they penalized the word 'women's' and downgraded women's-college graduates. The team found the bias, patched the named terms, and concluded the patch could not guarantee neutrality against unknown proxies — the model had learned the pattern, not the words — and the project was scrapped. The corpus's cleanest case of an organization detecting its own model's bias and choosing abandonment.
Explore this deployment in the PAN Lab →Workday, an applicant-tracking platform whose AI screening features operate inside thousands of employers' hiring pipelines at once, under a live federal collective action testing whether the vendor is directly liable as the employers' agent. On the litigation record the court sustained the agent theory at dismissal (2024) and preliminarily certified a nationwide age-discrimination collective (2025); a 2026 discovery ruling held the vendor's bias-testing privileged — the testing exists and is legally unreachable. Nothing here is a finding of discrimination: these are allegations and procedural rulings, not a verdict.
Explore this deployment in the PAN Lab →Unilever's graduate-hiring pipeline chained a games-based assessment with HireVue's automated video-interview scoring; the deployer reports ~90% faster time-to-hire, ~50,000 interview hours saved, ~£1M annual savings, and a 16% diversity improvement — every figure company- or vendor-reported, none independently audited. What makes it the domain's two-sided case is that both vendors' audit machinery is public in its honest but partial form: a cooperative academic audit of the games vendor, and the video vendor retiring its facial-analysis input under scrutiny. The blind spot: rejected candidates never re-enter the outcome data, so the effects are measured on hires only.
Explore this deployment in the PAN Lab →System map
Governance
For the actions behind these questions, see the Practice Library.
Seeing your organization in this domain? Mapping its actual pathways, pressures, and correction capacity is engagement work.
Work With 1023AIClaims made on this page and what supports them. The full registry lives in Evidence.
The volume's governance chapter places two standing duties on an agency that deploys AI, both distinct from any pre-deployment approval. First, an incident-reporting protocol that enables timely identification and remediation of algorithmic harm - discriminatory treatment, a biased risk assessment, a misdiagnosis, a breach of confidentiality. Second, transparent channels through which both the people served and the practitioners can report concerns or unexpected effects, so the accountability loop closes after deployment rather than ending at approval. The chapter is a conceptual synthesis and is cited as one: its four-tier social-work risk taxonomy is labeled by its own author as an original construction, informed by but not derived from binding regulation. It may be cited as a framework and must never be presented as a regulatory classification of any deployment in this registry.
Huang, J., Yang, F., & Lee, J. (2026). Ethical Challenges and AI Governance in Social Work. In R. An & M. A. Lindsey (Eds.), Artificial Intelligence in Social Work: Bridging Technology and Humanity. Springer. https://doi.org/10.1007/978-3-032-18443-6_22
doi.org/10.1007/978-3-032-18443-6_22
Appears in: AI in Social Work (Springer, 2026)
Topics: ai-ethics, ai-governance, social-work
