Business Schools Put Domain Expertise First And AI Second To Produce Hireable Graduates

Colleges spent the first years after ChatGPT trying to police it with detection software and syllabus policies written against its use. Both have held up poorly, since the tools struggle to identify AI-assisted work reliably and students reach for the technology regardless. As enforcement loses ground, what it was meant to protect becomes the harder problem. A degree is a claim that a graduate can do something, and that claim has to survive a job market where the same tools sit on every desk.

Chris Tidrick is Chief Information Officer at Gies College of Business at the University of Illinois Urbana-Champaign, a role he has held since 2023 that places him on the dean's executive leadership team. He has spent most of his career in higher education IT and his work covers technology and data strategy for the college, coordination with the wider university and outside partners, and the policy that governs both. He also serves as ex-officio chair of the college's technology advisory committee.

"A student still needs to develop their own domain expertise. If they develop that domain expertise and then understand how to leverage AI on top of that, then we have a superpower in the making. That is someone who is going to be able to leave the college with a degree, go out into the business world, and be even more successful than their peer who just has domain knowledge," said Tidrick. In his view, expertise has to come first, before a student adds AI on top of it. A graduate who knows the tools without knowing the field can't judge whether an answer is any good, or tell the system what to do differently.

• Literate, not dependent: Employers hiring business graduates are still working out where AI belongs in their own operations. Tidrick said many companies committed heavily on the expectation of a transformation that didn't arrive on schedule. What those employers ask the college for is graduates who can tell where the technology improves a process or a product cycle and where it adds nothing. "They certainly want our students to be AI literate, proficient, but not dependent," Tidrick noted. "You want them to be able to use the tools just like any other that we've ever had, that we've trained our graduates on."

• Assessment that holds up: Students will use AI on their work, and Tidrick treats policing it as a lost cause. The problem shows up in grading, where a take-home essay or problem set earns the same score whether a student reasoned through it or prompted their way to it. "A lot of the old evaluation mechanisms are not AI resilient at all," Tidrick explained. "Our traditional methods of assessment are going to deem them competent when in fact they're not." The risk compounds at the institutional level, since a degree is a claim the college makes about a graduate, and employers who find that claim unreliable stop trusting the rest.

Tidrick's team began working the question in spring 2023, a few months after ChatGPT reached general use, and the early hope was that higher education might finally standardize how it adopted a major technology. The pace of change ruled that out. The campus now runs ChatGPT Edu with institutional data protections applied, though that enterprise version doesn't carry the frontier models faculty want for their own work.

Faculty who want those models go outside the vetted set to get them. What comes out of that work stays with the person who built it, running on no shared infrastructure and covered by no process for turning one experiment into something the college supports. "The challenge for us is really creating an environment where people can safely experiment, having some guidelines around what data to put into AI and what data not to put into AI, which tools have been vetted and which tools haven't," Tidrick said.

• Strategy with a shelf life: A conventional technology strategy assumes the tools it names will still be the right ones a year later. "I could spend the next month developing the best strategy for AI for the Gies College of Business or the University of Illinois," Tidrick said. "The second we launch that new strategy, it will be outdated because the technology is so rapidly changing." He works instead from a set of principles that outlast any particular tool, on the expectation that the underlying technology changes about every six months. He also avoids commitments that would make switching vendors impractical later.

Tidrick applies the same standard to his own work. He runs drafts of reports and strategic plans through the tools, asking what a reader would find unexplained and which assumptions he has left buried. The questions it raises change the document before anyone else reads it, and he wants the technology working the same way for the rest of the college. "I use it very often as a thought partner, where I have some custom GPTs that are set up specifically to argue with me," Tidrick noted. "If I'm really set on an idea, I'll put that idea in and it will present all the alternatives to me and make me really think through what I'm doing."

• Workforce multiplier: Gies launched its online iMBA in 2016. Poets&Quants named it MBA Program of the Year for 2022, and it now ranks among the largest online MBA programs in the world. By Tidrick's account, the college has tripled its number of learners since that launch. Serving them has meant roughly 80% more staff and 50% more faculty, and his own team has grown from eight people to 38 over a decade. He put people at more than 75% of the college's expenses, so matching each increase in enrollment with new hiring runs into the budget quickly. "We are not intent on reducing workforce through AI, but we are intending on figuring out can it be a workforce multiplier for our existing faculty and staff," Tidrick said. "They can do more and we can reach more learners and have more impact with the same number of people."

What a degree certifies and how many people it takes to teach a growing student body turn out to be the same question, and Tidrick doesn't expect a quick answer to it. Companies facing the same question have to show results by the next earnings call. Universities work on longer timelines and can afford to get an answer wrong before they get it right. "I think higher ed is the place to figure this out because we are really well suited to figuring out complex problems, and this might be the most complex problem society has faced in quite a while," Tidrick concluded.