Governing the Digital Workforce: Agentic AI and the Future of Higher Education
I want to thank Sean Martin for challenging our thinking and institutional practices around agentic AI (https://www.youtube.com/watch?v=uZ-xzB5c_p8). While his discussion did not specifically address higher education, it is a timely issue, as AI agents become more capable, colleges and universities must determine who is responsible for supervising systems that increasingly perform work once assigned to people.
Agentic AI will compel institutions to confront a consequential governance question, who is accountable for AI systems that can act, decide, interact with data, and complete tasks across the enterprise? AI agents are no longer merely passive tools waiting for user prompts. They can be assigned roles, granted access to systems, evaluated based on outputs, and scaled across functional areas. In this sense, agentic AI begins to resemble a digital workforce rather than conventional software.
In higher education, this shift creates both opportunities and risks. Colleges and universities will need far greater visibility into where AI agents are being deployed across admissions, advising, student support, academic operations, IT, cybersecurity, finance, research administration, accessibility services, and teaching support. The challenge is that most institutions are designed to manage people through human resources and manage technology through procurement, IT, risk, and compliance functions. Agentic AI does not fit neatly into either structure. It performs work, interacts with data, influences decisions, and may affect students, faculty, staff, and institutional operations, yet it is not an employee in the traditional sense.
This raises new governance questions for institutional leaders. Who owns each AI agent? Who authorizes its use? What institutional data may it access? How is its performance evaluated? How are errors detected, reported, and corrected? Who is accountable when an AI agent makes a flawed recommendation, mishandles a student issue, produces biased output, or acts beyond its intended scope? Which decisions must remain human-led, particularly when they affect academic standing, financial aid, student records, accessibility accommodations, hiring, cybersecurity response, or learner success?
These questions are especially urgent in higher education because institutional trust depends on accuracy, fairness, transparency, privacy, academic integrity, and responsible stewardship. A poorly governed AI agent could affect the quality of advising, expose sensitive student information, generate inconsistent guidance, weaken compliance with accessibility requirements, or blur responsibility for consequential academic and administrative decisions. Therefore, agentic AI governance must extend beyond technical deployment. It must include role definition, ownership, data permissions, monitoring, auditability, compliance, cost management, value measurement, and clear human accountability.
The larger point is that agentic AI should not be treated merely as another technology initiative. It is an institutional leadership issue. Colleges and universities will need governance models that recognize AI agents as operational actors within the enterprise. The institutions that succeed will be those that establish oversight, accountability, and human-in-the-loop controls before AI agents become deeply embedded in core academic and administrative functions.
Ultimately, agentic AI has the potential to enhance institutional productivity, learner support, and operational responsiveness. However, that potential will only be realized if colleges and universities manage these systems with the same seriousness they bring to workforce planning, academic quality, cybersecurity, privacy, and enterprise risk. The question is no longer whether AI agents can perform work. The more important question is whether institutions are prepared to govern the work they perform.
