98 Percent Built AI Committees. 5.7 Percent Included Career Services

What 53 American universities built in response to generative AI.

98 percent of 53 American universities formed AI governance committees. 92 percent revised academic integrity codes. 85 percent provisioned enterprise AI tools. 7.5 percent documented employer outreach about AI expectations for graduates. 5.7 percent included career services in their AI governance structures. One institution out of 53 originated its AI strategy from employer input. The ratio across the dataset is roughly thirteen to one.

Why this happened

American higher education routed generative AI through the institutional channels that already existed: academic integrity, faculty governance, IT procurement, accreditation compliance. The channels routed AI to the questions those channels were built to answer. They produced governance. They did not produce readiness. The structures performed exactly as their theory of operation predicted. Organizational theory predicted this in advance: institutional isomorphism, loose coupling, the garbage-can routing of a new problem through the apparatus already in place.

The deeper finding

Beneath the governance-versus-readiness gap sits a measurement problem the governance layer cannot reach.

The essay, the problem set, the take-home exam, the term paper, the lab report, and the coding assignment were each built as measurement instruments for cognitive operations a student could perform alone. Generative AI can produce most of those artifacts at a quality that passes the reading. The instrument has lost its correlation with the trait. Governance levers (syllabus templates, integrity codes, detection contracts) sit a layer above the instrument. They cannot reach it. Integrity produces enforcement. Measurement requires redesign of the instrument and the rubric. The 53-institution data shows overwhelming investment in the first response and almost none in the second.

The outlier

One institution out of 53 built its AI strategy from employer input backward. Lehigh University placed its career services unit at the table where the institution set AI strategy. The Center for Career and Professional Development conducted fifteen structured employer interviews in summer 2025, organized the findings, and brought them to the provost's office. The Lehigh AI Fluency Framework and the three-tier AI Readiness program both trace their origin to that input.

Every other institution in the dataset has a career center with equivalent analytical capacity. The structural difference at Lehigh is procedural: the unit has standing in the channel where the institution decides. The seat is the variable.

What can move

The asymmetry is the residue of institutional decisions each governing body is positioned to revisit. Whether the institution's AI decision table includes the position from which the labor-market signal enters is a question accreditors, boards, governance committees, and career services units can answer for their own institution. The ratio is an outcome the field chose. The variable is one the field can reposition.

—

Hand-coded across 53 institutions, October 2025 through March 2026, ten-section public-source framework. Full paper 28,400 words. Carries the four-constituency walk through the gap, the AAC&U foil cluster reading, the Roche correspondence on what public-source coding cannot see, the Barkhi and Seref Journal of Macromarketing convergence from inside business school administration, and the Lehigh case at the variable where the break occurred. Working Paper V6.2, April 2026. Concept DOI 10.5281/zenodo.19647273.