Key points:
- Governance distributes responsibility, but does not diffuse accountability
- AI as a scaffold for learning: From access to judgment
- Beyond compliance: Governing higher ed in the age of intelligent systems
- For more news on AI governance, visit eCN’s AI in Education hub
Artificial intelligence is changing the conditions under which institutional purpose is interpreted, pursued, and realized through the exercise of institutional judgment. As institutions increasingly adopt AI, discussions related to governance have expanded to include fairness, bias, transparency, explainability, privacy, accountability, and regulatory compliance. These conversations are essential—yet, in focusing on how AI should be developed, deployed, and governed, we risk overlooking a more fundamental institutional question.
As intelligent systems increasingly participate in operational decisions, the defining challenge is no longer simply whether AI systems are developed, deployed, and governed responsibly, but whether institutions intentionally retain the capacity to ensure that purpose, and relevance, rather than optimization, continue to determine institutional direction.
Who is ultimately responsible?
Questions raised in leadership and governance board meetings increasingly focus on whether an AI system is fair, explainable, transparent, or free from bias, whether it complies with regulations, whether its outputs can be trusted, and whether appropriate safeguards have been implemented. While these are important questions, they focus primarily on the characteristics of the systems being deployed rather than on the leadership responsible for how those systems are used and shape the institution. The more important question is not whether intelligent systems can be trusted but who remains accountable for institutional direction.
Most institutions have responded by establishing AI governance committees, ethics panels, advisory groups, and cross-functional working groups. Each provides valuable expertise. None, however, owns institutional purpose in the same way as leadership does, because governance exists to ensure that institutional evolution remains a matter of deliberate choice rather than accumulated consequence. That distinction lies at the heart of governance.
Governance distributes responsibility, but does not diffuse accountability, nor does it diminish responsibility for institutional direction. While responsibility for AI may appropriately be shared among faculty, academic leaders, information technology, legal counsel, procurement, institutional research, and administrative units, and authority for implementation may be distributed across multiple levels of the institution, accountability for institutional direction cannot be delegated in the same way. Someone must retain both the authority and the obligation to ask a question that no intelligent system, committee, policy, or governance framework can answer: “Is this still the institution we intend to become?”
That question cannot be answered by an algorithm, or by compliance, assurance, ethics, or even through conventional governance structures. It can only be answered by leadership. Institutions can develop sophisticated policies, establish comprehensive governance structures, implement robust technical controls, and satisfy every regulatory expectation, and still gradually drift from their mission if no one remains explicitly accountable for questioning the cumulative direction created by hundreds of individually reasonable AI-supported decisions. The challenge is therefore not identifying who manages individual AI systems, but rather in ensuring that leadership remains accountable for the institutional direction emerging from their collective use. That is the responsibility leadership cannot delegate.
What happens when judgment is delegated?
Artificial intelligence is different from prior technological advances because in addition to increased efficiency it promises better recommendations, more consistent decisions, faster analysis, and in even superior performance for specific tasks. Used appropriately, these capabilities have the potential to improve decision-making. The risk, however, is not that AI begins making decisions but that institutions gradually stop exercising judgment.
There is an important distinction between automation and delegation. Automation removes repetitive work while allowing people to focus on activities requiring greater insight and expertise. Delegation occurs when organizations begin to accept recommendations without sufficient challenge, allowing optimization to substitute for deliberation, gradually transferring responsibility for judgment to systems designed to support, rather than replace, human decision-making. In such cases leadership rarely relinquishes institutional judgment deliberately but surrenders it gradually through hundreds of individually reasonable decisions.
The acceptance of an optimization that improves efficiency, a recommendation that consistently appears accurate, a predictive model that forecasts outcomes better than previous approaches or a workflow that reduces cost and saves time, is rational and beneficial. Yet, taken together, they gradually influence how priorities are established, resources are allocated, success is measured, and how institutional purpose is interpreted in practice. The human remains in the process but increasingly serves to confirm rather than critically evaluate the outcome. Judgment becomes procedural rather than intentional.
This is neither an argument against AI, nor for keeping people involved simply for the sake of human involvement. The question is whether institutional leadership continues to exercise the kind of judgment that cannot be delegated.
Judgment requires context, balancing competing priorities, recognizing when circumstances have changed, questioning assumptions, understanding institutional culture, and making decisions that extend beyond what legacy. These are not simply technical decisions but are acts of stewardship and leadership. This changes the nature of institutional decision making.
With AI, leadership could gradually shift from asking, “ What decision best advances our purpose?” to “Why would we override the recommendation?” subtly reversing the burden of proof. The distinction is subtle, but profound since the former begins with purpose while the latter begins with optimization. Recommendation becomes expectation unless leadership intentionally interrupts the process through independent action. Oversight becomes confirmation. This is precisely why governance can never be reduced to policies, controls, or oversight structures. Its deepest purpose is to preserve the institution’s capacity to think critically about itself, to challenge its own assumptions, and to intervene when optimization begins to diverge from mission.
Leadership remains accountable not simply for individual decisions, but for the cumulative institutional direction created by thousands of AI-informed decisions made over time.
Judgment therefore should be exercised not only in deciding whether a recommendation is technically correct, but in continually asking whether the objectives being optimized remain the right ones. The defining challenge is not whether intelligent systems can generate increasingly sophisticated recommendations, but rather whether institutions retain the discipline to decide when those recommendations should, or should not, shape their future.
What must leadership preserve?
When access to information was limited and specialized knowledge was scarce, institutional decision-making depended upon the ability of individuals and committees to gather, interpret, debate, and synthesize available evidence before acting. AI changes that environment fundamentally with institutional leaders having access to continuously updated intelligence capable of synthesizing knowledge, identifying patterns, evaluating alternatives, modeling scenarios, and generating recommendations at a scale and speed that would previously have been impossible.
While AI fundamentally changes how decisions are informed, it does not change who remains accountable for them. Instead, it raises the standard by which leadership should be judged since contrary to frequent predictions, AI does not diminish the importance of expertise but rather increases it. As intelligence becomes increasingly abundant, expertise becomes less about access to information and more about interpretation, integration, critique, and judgment. This distinction is particularly important in higher education where leaders must possess more than organizational capability, political acumen, or financial discipline. They must also understand the academic enterprise itself and its responsibility to the socioeconomic wellbeing of the region and nation, and thus possess the judgment to balance educational, scholarly, societal, and operational responsibilities while simultaneously understanding the context of workplace needs.
Higher education unquestionably must become more agile, more adaptable, more effective, and more relevant. Yet universities cannot simply become better managed corporations. Their purpose is not merely to improve institutional performance, but to strengthen the long-term intellectual, economic, scientific, civic, and workforce capacity of the region and nation. Success therefore cannot be measured solely by operational efficiency, enrollment, graduation, or employment outcomes. Strategic leadership must continually ask whether institutional priorities remain aligned with the broader public purposes for which higher education exists and whether institutional performance is being measured against the right objectives.
The defining responsibility of leadership is not simply operational oversight, but in ensuring that institutional evolution remains intentional, distinguishing optimization from purpose, efficiency from educational value, recommendation from strategy, and short-term metrics from long-term institutional relevance. It must recognize when the cumulative effect of individual rational decisions begins to alter the character of the institution itself. That demands judgment and courage to question increasingly persuasive recommendations, to challenge assumptions that appear self-evident, to intervene when institutional priorities begin to drift, and to make decisions whose value may not be immediately measurable but remain essential to the institution’s mission. And this applies both with and without AI.
Institutions rarely lose their way because leaders fail to understand the evidence before them.
More often, they lose their way because no one is willing to interrupt the momentum created by mass opinion, or optimization, when it no longer serves institutional purpose. Leadership is ultimately responsible for ensuring that increasingly abundant intelligence strengthens institutional judgment rather than gradually replacing it.
Is governance evolving quickly enough?
Artificial intelligence is changing not only how institutions make decisions, but the pace at which institutional direction itself can shift. Recommendations are generated continuously, workflows adapt more rapidly, and decisions that once unfolded over weeks or months can now be operationalized in hours or days, resulting in faster institutional movement, creating a governance challenge for which many universities remain poorly prepared.
Higher education has traditionally relied on governance processes developed for a more episodic operating environment, where decisions moved through sequential review, committee deliberation, and periodic approval before their effects became embedded across the institution. Governance designed for periodic decisions is increasingly being asked to guide institutions whose activities are continuously influenced by intelligent systems.
The answer is neither endless deliberation nor governance driven solely by operational speed. The former allows institutional change to outpace accountability while the latter sacrifices judgment precisely when most needed. Governance must become both intentional and agile to provide timely direction and accountability without sacrificing the deliberation required to protect institutional purpose. That requires a shift from governance as a sequence of approvals to governance as a continuing leadership function which provides ongoing visibility, timely accountability, and the capacity to intervene while institutional direction is still being shaped rather than after it has already changed.
Approval marks the beginning of governance rather than its conclusion.
Every governing board, executive leadership team, system office, and state agency should therefore be able to answer four fundamental questions:
- Who remains accountable for institutional direction as intelligent systems become embedded across the institution?
- Where is independent institutional judgment deliberately exercised rather than simply assumed?
- How does leadership determine whether the objectives being optimized continue to reflect relevant institutional priorities?
- How will leadership recognize when individually reasonable AI supported decisions are collectively changing institutional direction in ways that were never intended?
The challenge is not choosing between agility and deliberation, but in ensuring that governance is capable of both. Governance must therefore provide continuous strategic direction in an environment where institutional change is itself becoming continuous.
The defining responsibility of governance is therefore not to resist, or slow, institutional change but to ensure that it remains deliberate, accountable, and aligned with institutional purpose rather than becoming the unintended consequence of increasingly optimized decisions.
Vistasp M. Karbhari is a Professor in the Departments of Civil Engineering, and Mechanical & Aerospace Engineering at the University of Texas at Arlington, where he served as President from 2013-2020.