The image of a student revolt against artificial intelligence took center stage at last spring's graduation ceremonies, where boos and walkouts made headlines. But new data tells a different story. At American University's Kogod School of Business, regular AI use has quietly become the norm, and the loudest demand from students is not to ban the technology but to teach it more effectively.
These findings come from a three-year survey conducted at Kogod, which was shared for the first time with Axios. The survey tracked how business students have integrated AI into their academic and professional lives, revealing a dramatic shift in both frequency and purpose. In 2024, only 6.2% of students reported using AI eleven or more times per week. This year, that figure jumped to 29%. More than 80% of students surveyed said they had used AI for coursework within the past six months, cementing the technology as a staple of modern education.
Employers Are Pulling Students Toward AI
One of the strongest forces driving this adoption is the job market. The survey found that the percentage of students who faced AI-related questions in job interviews soared from 11.6% in 2024 to 42.6% this year. Students have taken notice. Prompt engineering has become the top skill they want taught in their business programs, while coding has leapt into second place, climbing from under 4% to nearly 20% in just one year. This is not casual experimentation; it is preparation for a workforce that increasingly expects AI fluency.
The shift is visible across student behavior. What began as a tool for occasional brainstorming or drafting has become a daily workhorse. Students are not merely checking whether AI can write an email or summarize an article; they are learning how to structure prompts, evaluate outputs, and integrate AI into complex problem-solving. The demand for prompt engineering reflects a broader recognition that knowing how to communicate with AI models is a distinct and valuable professional skill.
Tool Preferences by Age Group
The survey also uncovered generational and usage patterns in the choice of AI tools. Undergraduate students overwhelmingly prefer ChatGPT, while graduate students tend to lean toward Claude, which is known for its nuanced and careful responses. Interestingly, Perplexity topped raw usage numbers, but that statistic is flattered by a campus deal that gave every student free access to Perplexity Pro. When students were asked which factors mattered most in choosing an AI tool, accuracy ranked first, ahead of speed or cost.
These preferences may reflect different stages of academic development. Undergraduates often need broad assistance with foundational concepts, while graduate students are more likely to require deep analysis, critical evaluation, and sophisticated writing support. Claude's emphasis on safety and measured language may appeal to older students working on complex research projects. Perplexity's real-time search capabilities also make it a tempting choice for fact-checking, even if its high usage numbers are inflated by the free access.
The Anxiety Underneath the Adoption
None of this means students are uncritical about AI. Nearly half of Kogod students said they worried about academic integrity when using generative tools. In a separate survey conducted by Inside Higher Ed, 40% of students expressed fears that AI was eroding their ability to think for themselves. These concerns are real and persistent, but they do not translate into calls for rejection. Instead, students are asking for clearer guidance on how to use AI responsibly.
The ambiguity surrounding AI use has created anxiety for both students and faculty. Many students are unsure whether they are allowed to use AI for homework, exams, or even brainstorming sessions. They want explicit rules that distinguish between acceptable assistance and plagiarism. The survey suggests that educators who ignore AI altogether or try to ban it outright are missing an opportunity to teach students how to navigate a world where AI is ubiquitous.
The Paradox of Brainstorming
There is a paradox buried in the numbers. Brainstorming is the single most common use of AI, and it has held that position for three consecutive years. Yet research from Wharton School researchers found that people generate a wider range of ideas when they rely on their own thoughts combined with standard web searches than they do when they lean on ChatGPT. The most popular use of AI may therefore be the one that quietly narrows thinking, encouraging students to settle for the first plausible answer rather than push toward more creative solutions.
This finding raises important questions about cognitive offloading. If students use AI as a crutch during the idea-generation phase, they may miss out on the mental struggle that often leads to original insights. Brainstorming is not just about producing a list of ideas; it is about training the mind to make unexpected connections, challenge assumptions, and think across domains. AI can short-circuit that process, offering polished answers that feel complete but lack depth.
Some educators argue that the solution is not to ban AI from brainstorming but to teach students how to use it as a springboard rather than a finishing point. For example, a student might ask AI for a rough list of topics, then spend time discarding, combining, and refining those ideas before generating new ones. The struggle and synthesis remain human tasks, while AI serves as a starting point.
Schools Are Still Working It Out
Universities are lurching between extremes as they try to adapt. Kogod's interim dean, Casey Evans, explained that faculty first attempted to make assignments “AI-proof,” but quickly gave up on that approach. The new strategy is blunter and more practical. Staff members drop a student's work into an AI tool, produce a sample output, and show the student that their original effort was only worth a C. Then they coach the student toward an A by explaining what is missing and how to improve. Evans said resistance to this method has been rare.
Elsewhere, the mood is very different. Some universities have handed AI the responsibility of proctoring exams, using cameras and algorithms to monitor students for signs of cheating. Others, including Brown University, have retreated to in-person, handwritten papers to ensure that students produce original work. Detection tools and cheating tools keep chasing each other in an endless arms race. AI detectors flag perfectly legitimate text as suspicious, while students use increasingly clever prompts to bypass filters. The classroom has not settled on an answer.
This institutional chaos highlights a deeper uncertainty about what education should become in the age of AI. Some professors see AI as an inevitable collaborator, akin to the calculator or spell-check. Others see it as a existential threat to critical thinking and intellectual independence. The most forward-looking institutions are beginning to design curricula that treat AI as a subject of study in its own right, rather than simply banning or ignoring it.
The Bigger Picture and Its Caveats
The Kogod numbers carry an important caveat. The data is self-reported, drawn from 483 students at a single business school. It cannot claim to represent all college students across the country. But larger surveys point in the same direction. Inside Higher Ed found that 85% of students had used AI for coursework, while Gallup put weekly use at 57%. These are not fringe numbers. AI has become a mainstream academic tool, even if its use remains uneven and often underground.
The graduation boos were real. Some students and parents genuinely fear that AI is degrading the value of a degree. But the adaptation is equally real. Students walking into an AI-shaped job market may not love every aspect of the technology, but they are preparing for it. They are asking their schools to help them become proficient, ethical, and critical users of AI. The revolt, it turns out, was mostly a myth. The real revolution is happening quietly in classrooms, job interviews, and study sessions across the country.