I teach socially responsible finance at two universities -- one public and inclusive, the other private and selective. Before AI, my course taught students how to think for themselves, not just memorize facts.

But AI has decimated my teaching toolkit, and I can no longer offer students the opportunity to learn independently.

My case is not unique. AI has made it impossible for any teacher to assign research and writing outside the classroom. If this becomes the norm in education, we will damage our students irreparably – and the damage will spread to the society they will inherit.

We can’t let that happen. But countering AI requires major reinforcement of our education infrastructure. The costs will be substantial, but the alternative is unthinkable.

My story is a prime example of AI’s existential threat to education.

Here it is:

I used to require a research paper in my course – a short but intense exercise in applying the principles that the lectures presented. Students would approach me with ideas, I would help refine them into workable theses, and they would go off to explore.

The papers dealt with important questions for our day. Can we believe corporate icons like Walmart and Apple when they claim to be “sustainable”? Do capital markets reward (or punish) companies for their environmental behavior? Are business models like “Fair Trade” financially viable?

The students fretted over the reports, of course, but willingly accepted opportunities to think unconventionally, and to examine varied sources critically. They emerged with a more skeptical view of received wisdom, and sharper analytical tools to dig for the truth.

Two of the papers were so insightful that I collaborated with the students to convert them into articles that we published together (here and here).

Something changed a couple of years ago, as AI bots were improving dramatically and access was spreading rapidly. A number of the research papers were well-structured but had an unfamiliar and impersonal tone. I knew each student’s writing style, since they took written quizzes every week in class. This was not what I expected.

I considered trying to prove they had used AI, in violation of course guidelines. But that would have been time-consuming and futile, since AI detectors are fallible. And if I had accused someone in error, that would have been a grave injustice.

The next semester, I tried a new tack. I dropped the research paper, with regret, but found an alternative way for students to think creatively.

Attendance was mandatory, but students could make up for missing a class by consulting the lecture notes that were posted online after each session, then finding a related article from a reputable journal, and contrasting it to the notes.

Initially, it was clear that students were actively engaging with differing viewpoints. But over time, I began having doubts. The penny dropped when some students handed in papers for lectures that they had, in fact, attended. Apparently, they had carelessly fed the wrong notes into AI, then prompted it for a relevant reference and analysis. (I checked it out myself on a couple of AI apps, works perfectly.)

Consequently, now the course has no requirement for writing outside class. The students take in-class quizzes, midterm and final exam -- period. I don’t require oral presentations and debates because many students feel uncomfortable speaking in public, and the point of the class is to make financiers who act responsibly, not toastmasters.

Students generally give the course favorable evaluations, but they don’t know what they are missing. They could gain so much more from leveraging the course content into related fields, inventing original arguments, and building cases with independent investigation.

I have looked for ways to keep research skills alive (yes, even asked AI), but have not found ones that work.

Some sources claim that students use AI socratically, and learn from their dialogue with it. But I have found that students typically accept AI writing without question, even when submitting it in their own names.

Others advocate breaking the research paper into multiple steps and engaging with each student at each step, whether they use AI or not, to discuss and revise. But that would leave no time for the rest of the syllabus.

Research, analysis and composition are essential skills for incisive thinking and creative problem-solving. But today any research paper assignment can be fed into an AI app and completed in seconds. It is unfair, and unrealistic, to ask students to resist the temptation.

The only way to teach these skills going forward is under supervision and with AI turned off. Students will have to read texts and compose analyses during multiple, and lengthy,

proctored sessions. No longer can we expect students to spend time outside the classroom doing this work on their own.

This will put sizable burdens on both students and schools, but we have no choice. Students who can’t research and write critically will turn into scientists who can’t make discoveries, executives who can’t cope with change, and leaders who can’t envision the future.

Our students, and the world they will come to manage, deserve better.

Brad Swanson is an adjunct finance professor and author of “Profit vs. Progress: Why Socially Responsible Investment Doesn’t Work and How to Fix It” (MIT Press, 2026)