For decades, the basic bargain of education was straightforward: teach students something, put them in a room, give them a test and use the result to measure what they learned.
AI is making that bargain much harder to maintain.
This summer, exam controversies have erupted across several countries. In India, a massive exam-paper leak helped trigger nationwide protests and a political crisis. In Mexico, roughly 58,000 university applicants were ordered to retake an entrance exam after unusually high scores raised suspicions of widespread cheating. In Portugal, a failed attempt to digitize exam marking triggered protests from students, parents and teachers.
The details are different. The underlying problem is increasingly similar.
What exactly does an exam measure when technology can intervene at almost every stage?
AI Changed the Cheating Problem
Students have always cheated.
What has changed is the technology available to them.
Researchers cited by The Guardian describe a shift from traditional cheating toward sophisticated communication tools, including miniature cameras, earpieces, internet-connected devices and, increasingly, AI systems capable of supplying answers in real time.
That changes the economics of cheating.
A student no longer necessarily needs to know the answer, or even carry the answer into the examination room. They may be able to access someone, or something, that does.
Generative AI makes that problem even more difficult because it can produce essays, solve problems and explain concepts almost instantly.
The result is an uncomfortable question for educators: if a student can outsource the production of the answer, does the answer still demonstrate learning?
The Essay May Already Be Broken
That question extends beyond cheating during examinations.
AI can now produce a competent essay in seconds. That doesn't necessarily mean a student understands the material behind it.
Sarah Eaton, an academic-ethics researcher at the University of Calgary, argues that the traditional essay is increasingly becoming an ineffective way to demonstrate learning.
Her response has been to rethink assessment rather than simply search for better AI detectors.
One possibility is requiring students to defend their work verbally , effectively demonstrating that they understand what they submitted.
Denmark is moving in a similar direction, with teenagers required to give a verbal defense of written essays under new measures designed to address AI-assisted cheating.
That is a significant shift.
Instead of asking only "Did you produce this?", educators increasingly have to ask "Can you demonstrate that you understand it?"
The Bigger Problem Is Trust
This is where the recent exam controversies become more interesting than another round of headlines about students cheating with ChatGPT.
The real problem is institutional trust.
An examination only works if universities, employers and students believe its results mean something.
Once widespread cheating, leaked papers, unreliable marking or AI-generated work makes that assumption questionable, the damage extends far beyond individual students.
An exam grade is supposed to be a signal.
If nobody trusts the signal, the entire system starts to wobble.
That is what makes the developments in India, Mexico and Portugal worth watching. They demonstrate different ways in which the assessment system can fail, through human fraud, technology, institutional mistakes or some combination of the three.
AI Doesn't Have to Destroy Assessment
The answer probably isn't banning AI.
Students are going to use it. Universities are going to use it. Employers already are.
The more interesting question is whether education adapts its definition of learning.
If AI can write an essay, perhaps students need to explain the argument.
If AI can solve a problem, perhaps students need to demonstrate the reasoning.
If AI can generate a presentation, perhaps the student needs to defend the conclusions.
And if AI can answer a factual question instantly, perhaps memorizing the answer was never the most useful thing to test in the first place.
That could ultimately make education better.
But getting there requires institutions to redesign assessment rather than simply trying to catch students using increasingly capable technology.
What This Means for Miami
Miami has no shortage of universities, colleges and ambitious students, and the same assessment problem is coming here whether institutions are ready for it or not.
The question for Miami's education sector isn't simply how to prevent students from using AI.
It is whether universities can design assessments that remain meaningful when AI is available to everyone.
That could mean more oral examinations, live problem-solving, supervised work, project-based assessment and demonstrations of reasoning rather than simply submissions that can be generated outside the classroom.
The irony is that AI may force education to become more human, not less.
If software can produce the finished answer, the valuable part of the student's education may increasingly be the ability to explain why the answer is right, defend it, challenge it and apply it to a situation the machine hasn't already solved.
The exam isn't necessarily dead.
But the old exam may be.