Governing Human–AI Collaboration in Universities: Evidence Map and Capabilities

AI-mediated workflows now shape ordinary university work: feedback drafting, advising alerts, chatbot outreach, risk scoring, and case routing. The central issue is no longer adoption alone. When AI influences teaching, support, or administrative judgment, institutions must decide who may rely on the output, what review is required, and how students or staff can challenge an intervention that affects them.

This paper maps research on human-AI collaboration and governance in higher education from 2013 to 2025, with attention to generative AI after 2022. An auditable OpenAlex evidence-map workflow and seedless title/abstract triage trace where governance concerns concentrate and how the field describes AI's role. The synthesis finds a fast-growing discourse around surveillance/platformization, policy/ethics, advising, AI literacy, and teaching tools. It also identifies a framing gap: scholarship often treats AI as a transactional aid, while practice is entering care-intensive settings that depend on trust, consent, and recourse.

The paper proposes three governance capabilities: accountable delegation, epistemic stewardship, and consentful data stewardship with duty-of-care boundaries. These capabilities convert AI principles into routines that can be assigned, checked, and revised. Trustworthy human-AI collaboration depends less on tool choice than on institutional capacity to assign responsibility, protect evidence standards, and preserve consent and recourse.