AI Adoption at Higher Education Institutions: A Call for Designated AI Management

Synthesis: Kaže, Dombrovskis and Bolinskis asked 387 students and staff at higher education institutions in 13 countries who is responsible for AI governance at their institution, and found that almost nobody could say: only 17% knew of a designated AI-management responsibility. Of the rest, 71% thought one was necessary and 76% expected an AI manager to improve their experience. Five focus groups in Germany, Latvia, the Netherlands and Türkiye rated the same 15 competences as the survey and came back with a different profile — survey respondents pictured a compliance-oriented technical specialist, while the staff-dominated groups prioritized leadership, change and project management, and communication — so the authors' recommendation is designated AI management as change leadership with technical credibility rather than a technician post.

Key Findings

1. Designated responsibility is rare. Only 17% of 387 respondents knew of a designated AI-management responsibility at their institution, and the high share of "not sure" answers suggests knowledge of existing governance is itself thin.
2. Demand exceeds supply. Among respondents who did not report a designated responsibility, 71% considered one necessary and 76% expected an AI manager to improve their experience of AI at the institution.
3. The expected mandate is broad. Respondents placed training, ethical oversight, data security and strategy in the role, and split on whether it should be a dedicated position or a shared cross-functional function with an accountable lead.
4. Survey and focus groups produced different profiles for the same 15 competences. Survey respondents described a compliance-oriented technical specialist, while the staff-dominated focus groups prioritized leadership, change and project management, and communication alongside AI expertise and ethics.
5. Students are heavy users and were deliberately included. 56.3% of the sample were students (n = 218) and 78% of student respondents had already used AI tools in their studies, which is the authors' argument for consulting them about a governance role.
6. The sample is European and mostly private. 387 respondents from 13 countries, 51.7% from Germany and 64.9% affiliated with private institutions, with five focus groups and 41 participants completing the qualitative strand.

What the role is for

The paper's contribution is a tested definition rather than a survey finding alone. Starting from a literature-based nine-area competence profile developed in the Erasmus+ FORCE AI project, the authors kept the 15 competences that stakeholders could actually rate and rebuilt the profile around what the two methods agreed on: strategic oversight, ethical and legal alignment, data protection, staff training, and the soft competences that make any of those land in an institution.

That reframing matters because the survey and the focus groups disagreed in a systematic way. A questionnaire asking people to rate a hypothetical role pulls for abstract, compliance-type competences; staff who have watched change initiatives fail put leadership and communication first. The authors resolve the disagreement by treating the focus-group version as the more grounded account, which is a judgment about method rather than a result of the data.

Where the role should sit

Placement was the item respondents split on most evenly, between a dedicated post and a shared function. The paper's final position accepts both: designated AI management can be a single manager or a team, provided one person is accountable. For institutions already distributing AI policy across a CIO, a teaching center and a legal office, that is the practical form of the recommendation — accountability is the thing to designate, not necessarily a new hire.

The competence list was fixed in advance, which the authors name as a limitation: AI literacy as a distinct competence was proposed in a German focus group but could not appear in the ratings because it was not on the list.

What the evidence supports

The design is sequential mixed methods, but the two strands are not equally strong. The survey is a convenience sample with branching items, so the number of valid responses varies by question; country and institution type are almost completely confounded (German and Latvian respondents were predominantly at private institutions, Dutch and Turkish at public ones), so national and sector explanations cannot be separated.

The focus groups were fewer and smaller than planned in the Netherlands and in one Turkish session, and were staff-dominated. After correction for multiple comparisons, none of the student–staff differences on individual competences was statistically reliable, which the authors state plainly.

What this means for practice

• Name an accountable owner for AI, even without a new post. The finding is about designated responsibility, not headcount: a named lead inside an existing function satisfies the recommendation, and 17% awareness is the gap to close.
• Write the mandate wider than compliance. Training, ethics, data security and strategy were all expected of the role; a post scoped only to procurement or policy checking leaves most of the expected mandate unstaffed.
• Consult students before designing the role. Students were the heaviest AI users in the sample (78% had used AI tools in their studies) and are the group an AI manager's decisions reach first.
• Expect the job description to differ from what staff rate in a survey. The authors' contrast between hypothetical rating and focus-group discussion is a warning against building the post from questionnaire data alone.

Limitations

• Convenience sample, uneven by country and sector. 387 respondents from 13 countries, half from Germany, two-thirds from private institutions, with country and institution type confounded.
• Self-reported perceptions, including of existing governance. The high share of "not sure" answers means respondents' knowledge of their own institution's arrangements is limited.
• Hypothetical-role ratings. Competence items asked about a role most respondents had never seen in operation, which may favor abstract compliance competences over operational ones.
• The competence list was fixed in advance. Items absent from the list, such as AI literacy, could only surface qualitatively, and none of the student–staff competence differences survived multiple-comparison correction.
• Focus groups fewer and smaller than planned. The Dutch group and one Turkish session were undersized, and the groups were staff-dominated.
• No formal back-translation. The questionnaire was translated into four languages by review and consolidation rather than back-translation, which the authors note.

Citation

Kaže, V., Dombrovskis, V., & Bolinskis, G. (2026). AI Adoption at Higher Education Institutions: A Call for Designated AI Management. Education Sciences, 16(11), 1839.