In a new study, colleges and universities reported more progress in forming artificial intelligence (AI) committees and offering training than in dedicating staff, funding, and regular policy reviews to sustain that work.
D2L, the WICHE Cooperative for Educational Technologies (WCET), and Opened Culture announced the findings Oct. 21 in a press release. The resulting guide, Governing AI from Strength, draws on a survey and eight in-depth interviews to examine how academic leaders make AI governance decisions and develop the skills to carry them out.
Among 54 survey respondents, 41 reported an AI task force, steering committee, or advisory body at their institutions, but only 18 reported dedicated funding, 14 reported dedicated staffing, and nine reported policy review cycles.
The survey was open April 27 – Oct. 5. Respondents who provided location information represented nine countries, including 30 respondents in the United States. The sample included leaders from public universities, private institutions, community and technical colleges, and system offices; the authors cautioned that the findings do not represent higher education as a whole.
The research describes institutions making decisions about acceptable use, privacy, procurement, and student access, often across several offices but without clear ownership.
Institutions making progress connected existing policies and committees, assigned responsibility, and scheduled reviews.
“When it comes to AI, many of our members feel pressure to build something entirely new, and fast,” said Van. L. Davis, executive director of WCET. “What this research makes clear is that institutions don’t have to start from scratch. The governance structures, policies, and shared commitments they already have are real assets.”
The guide recommends that institutions establish a standing governance body with a clear charter, decision authority, and a renewal date. Its charter template asks leaders to specify which decisions the group can make, which require approval elsewhere, and how its work connects to faculty, technology, and curriculum governance.
Uneven understanding of AI among decision-makers emerged as the most frequently cited obstacle, raised by 22 respondents. According to the guide, institutions that acted on this concern required a short preparation course for governance participants, with senior leaders completing it first.
The authors recommend giving governance groups practical work, such as testing tools, running a limited pilot, or gathering student feedback, to build their understanding of AI.
Tool access presents another governance decision. The guide recommends documenting which AI tools the institution supports, who can use them, who pays, and what vendors commit to regarding data use, retention, and access. Its decision worksheet asks institutions to address accessibility, alternatives for people who decline to use AI, and dates for reviewing the arrangement.
Examples in the guide show how institutions incorporated those decisions into existing operations. An unnamed public university in North America added an algorithmic impact assessment alongside its privacy review during procurement. A U.S. community college district incorporated three AI statement options into its syllabus publishing system, allowing faculty to select a statement and add course-specific guidance.
The guide includes five worksheets and toolkits and a road map organized by actions for each month and semester, and annual review. For chief information officers, the roadmap recommends establishing a baseline of supported tools, coordinating procurement and implementation with academic and administrative partners, and reviewing annually who can and cannot access approved tools.
