College administrators are increasingly turning to AI chatbots as unofficial partners to accomplish daily tasks.
As emerging tools and heightened user sophistication unlock greater efficiency and insight, administrators must toe the line between personal experimentation and compliance with institutional policy.
This article is part one of a three-part series, “Navigating AI,” which explores how administrators are properly embracing AI use. Click here for part two.
“Administrators are consulting with chatbots more than many people realize—often quietly,” Dave Schippers, provost at Walsh College in Michigan, said in an email. “Many institutions are still building governance, policies and shared norms as they go.”
Chatbots cover a breadth of tasks for Katie Condon, assistant vice president of enrollment at the University of West Florida. She first used AI to navigate thousands of pages of government regulations and to identify trends in enrollment across various programs.
Condon now takes it one step further by transforming those data reports into captivating presentations for a more general audience. With the keen eye of an editor, AI recently helped Condon refine the university’s strategic enrollment plan, which was reviewed by the provost.
“We’re always asking where we can accomplish more with less in higher ed,” says Condon. “My use [of AI] has evolved,” and it has “given me time to work on strategy instead of being stuck in the weeds.”
More on AI: Faculty worry there are several ways AI is harming students
Gregor Thuswaldner, provost of La Roche University in Pennsylvania, uses ChatGPT or Microsoft Copilot when crafting official communication with different campus community members.
One on level, AI helps Thuswaldner wade through “low-level emails” with which he’s regularly bombarded. With the click of a pre-saved prompt, he can quickly draft a response declining a vendor’s services.
Thuswaldner also uses chatbots to refine his communication with staff and students.
“I ask AI to poke holes in my argument and help me find my blind spots, so that I can refine my ideas,” he says. It helps find “a better grammatical structure or thought that’s slightly better expressed than perhaps my blunt way of saying it.”
“The important thing is that we remain in the driver’s seat,” Thuswaldner adds. “We’re not mindlessly outsourcing things to AI.”
Schippers, of Walsh College, consults with various customized chatbots to better inform his decision making. Connected to the institution’s data, each chatbot offers a perspective related to the college’s policy and regulations, operations and other stakeholders.
Such ability to quickly stress-test ideas to empower human judgement demands that the user “treat AI as an advisory force multiplier, not an authority,” Schippers says.
The limits—and risks—of using chatbots
Administrators across institutions can work better and faster with chatbots and AI-embedded systems, says Jena Zangs, chief data and analytics officer at the University of St. Thomas in Minnesota.
But administrators who over-rely on information that AI regurgitates from the Internet run the risk of burying their own ideas.
“Each one of us brings a different perspective and voice to the table, and I think that’s where AI can be intimidating,” she says. “How am I expecting my team to use AI while understanding its limitations and concerns around safety?”
Administrators must limit the information they provide to a chatbot—and know what tools they’re allowed to use. “Shadow use” occurs when administrators draft proprietary information with chatbots not sanctioned by their institution, according to Schippers from Walsh College.
“Administrators routinely paste more into chatbots than they would ever say out loud in a meeting because it feels like a private conversation,” he says.
AI prompts that contain the personal data of students or staff can trigger FERPA violations. Entering non-public documents, such as certain vendor contracts and institutional strategic decisions, can lead to a “loss of control.”
“You’ve now copied regulated or sensitive information into a separate system, with its own retention, audit, admin access and legal exposure rules,” Schippers says. “Even reputable platforms can face changing retention requirements due to litigation holds and court orders.”
Zangs from St. Thomas University has heard “scary stories” where unauthorized staff were able to access details from an executive-level meeting due to an AI note-taking assistant.
Institutions must provide clear guidance on which data cannot be used with AI, Schippers says. While chatbots can be helpful, they should not substitute judgment, policy or accountability.
“The decisions are mine and the consequences lie with me,” Schippers says.
Stay tuned for part two, where administrators share tips and tricks on how to responsibly use AI amid regulatory and safety concerns.
