FID-085 · Open question
Contestability and Recourse in AI-Mediated Public Services
Can people understand, challenge, and correct AI-mediated decisions that affect their access to public or community services?
Why the question remains open
An explanation is not meaningful recourse if no one can correct the decision or accept responsibility. Research should examine whether affected people can exercise agency, including those with limited time, language access, or technical familiarity.
Working hypothesis
A proposition to test, not a finding.
Evidence-linked explanations paired with a named human review route will improve successful error correction over explanation-only interfaces. They may fail when reviewers lack authority or access barriers remain.
Proposed method
How the question could be tested
- 01Create synthetic eligibility, referral, and resource-allocation scenarios with documented rules and deliberately seeded errors; use no real benefit decisions.
- 02Compare explanation-only, appeal-form, and accountable human-review workflows with consenting adult participants and service practitioners.
- 03Measure recognition of errors, ability to contest, correction accuracy, time and effort, accessibility barriers, and unresolved cases. Separate user comprehension from actual institutional correction.
Needed controls
What must constrain the study
- 01Obtain appropriate ethics review and consent. Avoid recruiting people in ways that imply participation affects services.
- 02Include affected-community input and language/accessibility needs. Disclose contested policy assumptions instead of presenting administrative rules as inherently legitimate.
- 03Ensure comparison reviewers have specified authority and matched evidence. Report burdens and failures, not only average completion rates.
Relationship to existing work
Extends FID-038's concerns about consequential profiling and FID-064's collective judgment questions to practical contestability and institutional correction.
Expected outputs
Artifacts the work should produce
- 01A contestability evaluation protocol and synthetic service scenarios.
- 02A design-and-governance report connecting explanations to enforceable correction routes.
Open questions
Uncertainties the protocol must resolve
- 01Who bears responsibility when a model, vendor, and institution share a workflow?
- 02When should a service decline automation even if its explanations are understandable?
Related calls
Continue through this research area
FID-069
Verifiable Delegation and Revocation in Multi-Agent Networks
How can people and institutions verify which human, organization, agent, or sub-agent is acting; what authority it received; what limits apply; and whether that authority has been narrowed or revoked across a multi-principal agent network?
FID-071
Confidential Agent Memory and Cross-Context Disclosure
How do persistent memory, summaries, retrieval stores, tool traces, delegation, and exports cause confidential context to influence or leak into unrelated sessions, roles, tasks, or organizations? Which technical controls make purpose limitation, deletion, and revocation testable?
FID-077
Independent Agent Incident Investigation and Evidence Sufficiency
What operational evidence lets independent investigators reconstruct an agent incident, distinguish competing explanations, and identify which interventions could have changed the outcome?
Open question
Open work
Primary need: public service expertise, community participation, governance, accessibility research
- Public-service and nonprofit practitioners to model realistic review routes.
- Affected-community representatives and accessibility researchers to define meaningful recourse.