Not yet human-reviewed
This source record was written automatically by the research pipeline and has not yet been reviewed by the maintainer or integrated into a topic page. Treat it as a provisional summary and verify claims against the original source.
Co-constructing sociotechnical AI governance: participatory system mapping using algorithm registers
Citation. de Troya, Í., Enbergs, M., Doorn, N., & Dobbe, R. (2026). Co-constructing sociotechnical AI governance: participatory system mapping using algorithm registers. arXiv:2608.12166.
Summary. This preprint examines whether a municipal algorithm register supports public accountability for a welfare-benefits eligibility decision-support system. Interviews, surveys, and participatory system-mapping workshops with municipal staff, civil-society organizations, and ombudsmen informed a system-theoretic process analysis. The authors find that the register alone obscured governance relationships and potential harms that stakeholder mapping made visible.
Key claims
- An algorithm register’s disclosure of risk assessments and governance instruments can be too abstract for external stakeholders to assess how a system is governed or whether safeguards are effective.
- Participatory mapping with direct and indirect stakeholders identified potential harms—including erroneous benefits denial, performance deterioration, and barriers to contesting wrongful decisions—that would not have been evident from the register alone.
- Governance documentation should situate technical systems within the wider actors, procedures, resources, complaint channels, and accountability relationships that shape system safety.
- Participatory mapping can itself support critical reflection and stakeholder scrutiny, while system maps still risk omitting affected citizens’ lived experience.
Evidence & limitations
- Qualitative single-case study of a Dutch municipal welfare-benefits eligibility tool; the reported participatory work involved municipal staff, civil-society organizations, and ombudsmen (N=8), not system developers or affected citizens directly.
- The paper is an arXiv preprint and its governance findings are context-specific rather than evidence that participatory mapping improves accountability across deployments.
- Public provenance was verified against the arXiv PDF for 2608.12166v1: it has the same SHA-256 as the ingested Drive artifact.