Human-AI Collaboration

Human-AI collaboration in public sector decision-making creates an "accountability black hole," where blurred responsibility leaves citizens unable to identify decision-makers. Existing research treats traceability as a retrospective auditing tool rather than a proactive design principle. Addressing this gap, the Traceable Collaboration Framework (TCF) reconstructs human-AI decisions as an evidence chain with four interconnected links: Value Traceability (policy intentions), Logic Traceability (algorithmic reasoning), Interaction Traceability (human discretion), and Audit Traceability (record integrity). Accountability depends on the integrity of the entire chain. Integrating chain-of-custody theory, street-level bureaucracy theory, and information quality research, the TCF diagnoses cases across the Netherlands, UK, US, and China. Theoretically, it elevates traceability to an epistemological concern, engaging with evidence law, XAI, and algorithmic accountability scholarship. Practically, it provides public managers with an Evidence Chain Integrity Checklist and Traceability Maturity Model. Ultimately, the framework establishes contestability—empowering citizens to genuinely challenge decisions—as the ultimate purpose of traceability. Keywords: Traceable Collaboration Framework; Algorithmic Accountability; Human-AI Collaboration; Digital Governance; Public Sector Algorithms; Chain of Evidence; Street-Level Bureaucracy; Contestability