What If Our AI Strategy Succeeds?

• Students often use AI not to avoid thinking, but to navigate uncertainty. In this case, faculty should make trust, rather than surveillance, the foundation of AI education.
• Business schools that center AI education on trust will assess students’ ability to question, override, and justify AI outputs—not simply document or restrict AI use.
• Training faculty, developing shared AI norms, and redesigning assessments around trust can help graduates build the judgment they will need to collaborate critically with AI.

 
When I first learned about generative artificial intelligence in late 2022, I opened class with a threat. I told my students I could easily detect AI‑written work—that was a lie delivered with the confidence of someone who had just earned the keys to academia. Then, trying to gauge the “enemy,” I asked how they were actually using these tools. One shrugged. '“I asked it what to have for breakfast.” Another, without irony: “I ask it what to wear.” We laughed.

My moment of clarity came months later, when I was in my office with a student whose work I had flagged for “suspiciously clean” prose. But when she pulled up her ChatGPT history, it did not include a single prompt related to her schoolwork. Instead, the prompts focused on helping her navigate her everyday life: “Explain this lease clause like I’m 19.” “Draft a text to my landlord about mold.” “Help me sound professional emailing a professor who hasn’t replied in two weeks.”

She wasn’t outsourcing her education. She was outsourcing her anxiety. I had spent a semester training myself to see evidence of fraud where I should have seen resourcefulness.

For business schools, this misdiagnosis is exceptionally costly. Consider AACSB’s 2025 report “GenAI Adoption in Business Schools: Deans and Faculty Respond,” which surveyed 236 deans and 429 faculty across more than 60 countries. In it, deans exhibit greater optimism about GenAI adoption, while faculty express more caution, citing ethical concerns and questions of readiness. Both groups agree on the importance of integrating GenAI into the curriculum, but that goal cannot be realized if faculty and students have different understandings of what the tool can do.

The data confirms that this divide is real and wide. A 2024 AACSB Insights article cites an internal survey at Vlerick Business School in Brussels. The survey revealed that 100 percent of master’s students and 87 percent of MBA students had used GenAI for research, brainstorming, or drafting, but only 71 percent of faculty reported doing the same. Students are using AI for lease clauses and landlord texts, then bringing that fluency to our assignments. We then see the polished result and assume deception.

Similarly, the Digital Education Council’s 2024 Global AI Student Survey, based on 3,839 responses across 16 countries, found that 86 percent of students regularly use tools such as ChatGPT in their studies. Yet only 34 percent felt their universities actively sought their feedback on AI matters. Such an expectation on their part is not defiance. It is rational self-advocacy in a vacuum of leadership.

The Academic Policeman

These differing assumptions shape not only how business schools write policy, but also how we teach. In our classrooms, many of us, as business faculty, are locked in a struggle of our own making. Fear of AI‑driven cheating has forced some of us into a punitive, surveillance‑oriented stance that fundamentally alters the teacher‑student relationship—pushing educators into what researchers have called the “academic policeman” role.

A significant minority simply opt out: A 2025 study found that more than one‑third of surveyed faculty abstained from using GenAI. They cited reasons such as “not ready/not now,” “no perceived value,” “identity in tension,” and “threat to human intelligence.” According to a recent article in The Chronicle of Higher Education, some are even retiring earlier than they intended rather than grapple with the impact of AI on their teaching.

Some research suggests that students are more concerned about the misuse of AI than we are. The recklessness is not in their prompts. It is in our silence.

When our classroom energy is spent hunting for machine‑generated prose, we become detectives, not mentors. Students sense this suspicion and retreat further into silence.

This stalemate happens when trust erodes and institutional policy remains fragmented. Jiahui Luo’s 2024 study found that the rise of GenAI is causing an “erosion of trust” between students and teachers. The research highlights an absence of “two‑way transparency”—students are required to declare their AI use, but their teachers often do not do the same. This results in a “low‑trust environment where students feel unsafe to freely explore GenAI use.”

Compounding this, the AACSB report on GenAI adoption revealed that while 47 percent of deans say their schools have implemented AI policies, many schools lack clear or actionable guidance, and nearly 45 percent lack the coordinated governance to translate policies into practice. As one recent analysis puts it, “the worst AI strategy in higher ed is no strategy at all.”

Here is the finding that should reframe every business school leader’s thinking: Some evidence suggests that students may be more reflective about appropriate AI use than faculty often assume. A 2025 study of Chinese university teachers and students found that “despite the differences in use patterns, both teachers and students exhibited similar positive attitudes toward the utility and positive impact of GenAI.” Crucially, the survey also found that “students seemed to be more concerned about the use of GenAI tools than teachers.”

The people we think need policing are more concerned about misuse than we are. The recklessness is not in their prompts. It is in our silence.

What If Our AI Strategy Succeeds?

Up to this point, many educators have focused on a singular question: What if our students use AI to cheat? Given the research cited above, let me pose a harder question: What if our current AI strategy succeeds in addressing our concerns about AI beyond our wildest expectations?

Imagine a business school in 2030. Its administrators include detailed AI disclosure requirements in every assessment. Its faculty screen every assignment through multiple verification systems. Every student completes mandatory AI ethics training, and every faculty member follows standardized institutional guidelines. Compliance rates approach 100 percent.

From an accreditation perspective, the school appears exemplary. Policies are clear. Processes are documented. Risks are controlled. Yet something unexpected has happened: Graduates become highly skilled at explaining how they used AI, but less confident in deciding when to ignore it. When faced with ambiguous situations, they search for procedural certainty before exercising judgment. When an AI recommendation conflicts with local market knowledge, stakeholder intuition, or ethical concerns, these graduates hesitate—not because they lack intelligence, but because years of educational conditioning have taught them that compliance matters more than challenge.

We did not build business schools to train for obedience. We should stop building AI policies that do.

Consider a newly hired marketing analyst asked to evaluate an AI‑generated customer segmentation strategy. The model recommends targeting a profitable demographic segment while excluding several smaller customer groups. The analyst recognizes that the recommendation may create reputational risks and overlook emerging market opportunities. Yet instead of questioning the output, she focuses on whether the AI process complied with organizational policy.

Or consider an MBA graduate serving on a team evaluating international expansion opportunities. AI‑generated analyses consistently rank one market as the preferred option, but local partners raise concerns that this recommendation overlooks political and cultural factors. The graduate knows the AI may be missing context but chooses to comply with guidelines rather than challenge machine‑generated conclusions. The result is not misconduct, but something potentially more dangerous: Overconfidence in procedural correctness.

These scenarios are not dystopian fiction. They are the logical endpoint of a compliance‑only AI strategy. But we did not build business schools to train for obedience. We should stop building AI policies that do.

What Winning Actually Looks Like

A trust‑based strategy asks a different question: What if our approach to AI integration succeeds in preparing students to collaborate critically with AI under real‑world pressure? If we want to move forward with this goal in mind, here are three shifts I would make if I could go back to that classroom in 2022—each one is a change I think every business school should adopt today:

1. Invest in faculty AI literacy. Most faculty identify as novice or intermediate GenAI users, according to AACSB’s 2025 survey. Given that, schools should create discipline‑specific workshops on prompt‑crafting, AI limitations, and ethical use. For instance, Vlerick Business School’s Learning Hub offers faculty the Assignment Profiler, a GPT that flags assignments completable solely by AI. This helps faculty redesign for human input.
2. Co‑create AI guidelines with students, not for them. When students help draft classroom norms, compliance becomes ownership. We should explore important questions regarding AI’s use with our students: What is acceptable collaboration with AI? What constitutes attribution? If I had co‑constructed these norms with my students from the start, I would have had partners in learning instead of suspects under surveillance.
3. Replace the academic policeman with the curious mentor. We need to redesign assessment around trust and reward process over product. Students should make oral defenses; write iterative drafts with visible revision history; and complete simulations where AI is a collaborator, not a replacement for critical thinking.

Through these assignments, students’ choices compound under pressure as they learn to defend their reasoning in real time. As described in AACSB’s A Framework for Artificial Intelligence in Business Education, “AI is used for hypothesis generation and data screening, but final interpretive tasks remain human‑only.”

We are seeing this shift in practice. At the Wharton School of the University of Pennsylvania in Philadelphia, for example, the Pincus AI Lab for Organizational Innovation explores “human‑AI co‑intelligence.” At the University of Newcastle in the U.K., faculty made AI use mandatory in innovation and entrepreneurship courses, then built programmatic assessment frameworks capturing how students produce work.

When Jared Harris, a professor at the University of Virginia Darden School of Business in Charlottesville, discovered that ChatGPT produced the same wrong answer that stumped students, he projected that answer on screen in class. He asked students, “How do you know if GenAI is feeding you something wrong?” Such moments led Darden to integrate AI into the core curriculum, teaching interrogation over acceptance.

Assurance of learning standards must evolve—from verifying that work is AI‑free to verifying that students can articulate when and why they override AI output.

At the University of Portsmouth in the U.K., I redesigned an MBA consultancy course around trust. In this pilot, 143 students in 29 teams worked with real small and medium-sized enterprises over eight weeks, using any AI tool at any stage. The requirement: Create a 10‑minute presentation explaining every time they used AI in a material way, every time they overrode its output, and why. One team overrode ChatGPT’s premium pricing after a client interview revealed local price sensitivity the tool could not know. Another caught two hallucinations by manual verification.

With this approach, students do not hide their use. They argue about it. They defend their judgment. Trust is not the enemy of rigor, but the precondition for the cognitive friction that produces real learning.

A Call for Accreditors (and for All of Us)

If the goal is to produce graduates who critically collaborate with AI, then assurance of learning standards must evolve—from verifying that work is AI‑free to verifying that students can articulate when and why they override AI output.

This argument raises a question for accreditors. AACSB has signaled openness, noting in its 2025 GenAI Adoption report that schools need “consistent and practical policies, alongside training on ethical concerns, to foster responsible GenAI use.” But we must go further. “Responsible use” cannot mean only compliant use. It must mean discerning use.

The question for accreditors and every business school leader is whether we will measure what is easy or what is essential. In the cheap‑answers era, the premium is not on producing answers, but on owning judgment, owning meaning, and owning responsibility, as Timothy Hor puts it in his 2026 AACSB Insights article. If our assurance systems do not capture that, they are assuring compliance, not learning.

The breakfast question I heard in 2022 was innocent, even trivial. But it contained the seed of something I was too threatened to see. Students were already talking to the machine, already experimenting, already building a relationship with the technology that would reshape their professional lives.

That means that our job is now to admit we got it wrong, and then to learn alongside them, starting with what they already know. If we succeed, our graduates will not need permission to challenge AI. They will have the discernment to do it naturally. And that is a future worth building.