Beyond AI Compliance
Five questions for meaningful human authority in the age of AI
This brief offers five questions for provosts and academic-affairs leaders examining what should remain meaningfully human as their institutions adopt AI. It supplements risk-oriented governance frameworks with an institutional-purpose test — Personhood, Answerability, Formation, Presence, and Efficiency — that returns each consequential practice to a single decision: Preserve, Redesign, or Release.
AI use has already outrun institutional guidance. In a 2026 EDUCAUSE survey of 1,960 higher-education employees, 94 percent reported using AI for work during the previous six months; 54 percent said they were aware of policies or guidelines governing that use.[2]
ReHuman is an interdisciplinary framework for the question policy leaves open: what should remain meaningfully human as universities adopt AI? It helps academic leaders examine personhood, answerability, formation, and presence, then decide what to Preserve, Redesign, or Release.
Throughout, consequential describes a decision or practice that materially affects a person's standing, access, development, reputation, or ability to contest an outcome.
Responsible AI governance manages system risk. ReHuman makes the institution name the human purpose the system must serve.
Where governance can still leave the wrong practice intact
Consider an AI-assisted advising system. The university protects student data, tests performance, documents use, monitors outcomes, assigns oversight, discloses the system's role, and gives advisers override authority. Those are serious governance controls, consistent with the operational discipline of NIST's Govern, Map, Measure, and Manage functions.[1]
Yet the adviser may still become a messenger for the system: delivering recommendations without interpreting a student's circumstances, exercising professional judgment, or helping the student develop judgment of their own. Because the AI RMF is voluntary, non-sector-specific, and use-case agnostic, it does not determine what academic advising is for. ReHuman supplies that institutional-purpose test.
| A governed implementation asks | The ReHuman test adds |
|---|---|
| Is the system trustworthy, monitored, documented, and subject to accountable human oversight? | Does the workflow require the adviser to document a contextual interpretation, exercise independent judgment, and own the recommendation — or can the system's output pass through unchanged? |
The same test applies when AI supports hiring review, tenure evaluation, research assessment, or other consequential institutional judgments.
The Monday-morning decision: Preserve · Redesign · Release
- Preserve.
- Direct adviser judgment when context, development, or meaningful contestability shapes the value of the encounter.
- Redesign.
- Use AI to prepare options and surface patterns while requiring the adviser to interpret, question, and own the recommendation.
- Release.
- Automate scheduling, reminders, routine summaries, and other clerical work where relational consequence is minimal.
Five institutional questions
Personhood — Does the institution see more than the profile?
The representation is not the person.
Context should be reviewed before a classification drives action; firsthand knowledge and information the system cannot capture should carry decisional weight.
Operational test. The affected person can explain their circumstances to someone with authority to reconsider the result.
Answerability — Who owns the consequential judgment?
An answer is not enough. Someone must remain answerable.
A named person should own the outcome, explain the reasoning, and possess authority to revise the decision.
Operational test. The decision record identifies the responsible reviewer and states what that person affirmed, revised, or rejected from the system's recommendation.
Formation — What capacity is this practice cultivating?
The artifact is not the formation.
Formation-based AI ethics directs attention to how repeated AI use shapes attention, judgment, responsibility, and agency.[4] Faculty concern is already substantial: in an Elon/AAC&U survey of 1,057 faculty, 95 percent anticipated increased student overreliance and 90 percent anticipated diminished critical thinking; the authors caution that the findings are not generalizable.[3]
Operational test. When reasoning, judgment, voice, or professional identity is a learning objective, students demonstrate that capacity at some stage independently of AI — through explanation, revision, oral defense, live problem-solving, supervised performance, or transfer.
Presence — When must a person participate directly?
Direct human participation is required when interpretation, relationship, moral judgment, or meaningful contestability is part of the value being delivered. This criterion generates the relevant cases: advising, mentoring, crisis response, discipline, appeal, and consequential evaluation.
Efficiency — What should be preserved, redesigned, or released?
Stewardship of Efficiency returns every consequential practice to the Preserve, Redesign, or Release decision above — and requires leaders to revisit that decision as the technology, evidence, and institutional context change.
The ReHuman Formation Index (RFI) can support this work as a pilot diagnostic across Personhood, Answerability, Formation, and Presence. It is designed as a conversation starter rather than a ranking, grade, or certification. Its purpose is to surface evidence, tensions, and one practical decision an institution can act on.
Request a briefing
Provosts and academic-affairs leaders can request a 30-minute conversation about applying the ReHuman test to one practice at their institution. A response will follow within two business days.
English