Here is a story about trust, told in three acts. In Mexico City, the UNAM is administering in-person control exams after detecting irregularities—including the possible use of artificial intelligence—in its online admissions test 1. In Arizona, a university is launching a degree to train influencers, teaching students to build a personal brand 6. And in Argentina, a new index shows that only half of students who started primary school in 2020 reached sixth grade on time with satisfactory skills in Language and Mathematics 4.
These are not separate stories. They are the same story about what happens when an institution’s incentives collide with reality.
Start with the classroom problem. The UNAM’s online exam was designed to be accessible and efficient, but it created an environment where the institution could not verify who was actually doing the work. The response—a return to proctored, in-person testing—is a logistical retreat from a technological advance. The incentive that produced this problem is simple: the university wanted the convenience of scale without the cost of verification. Online proctoring software exists, but it is expensive, invasive, and imperfect. So the institution chose the cheaper path and paid for it in credibility.
The practitioner evidence here is familiar to anyone who has taught. When assessment is unverifiable, the assessment becomes meaningless. Students know this. They will use the tools available to them, and if the institution leaves the door open, they will walk through it. The UNAM’s decision to re-test 736 aspirants in León alone 1 is not a punishment; it is an acknowledgment that the first test measured nothing.
Now consider the ASU influencer degree 6. The conventional reform narrative says this is a capitulation to the market—a university selling out to the attention economy. But the more interesting problem is the incentive structure. A degree in content creation is a degree in self-branding, which means the product being assessed is the student’s public persona. How do you verify that? How do you grade authenticity? The UNAM’s problem was that the test could be faked; ASU’s problem is that the subject itself is a performance.
The Argentine data 4 is the sobering backdrop. Half of students arriving at sixth grade without expected skills is not a technology problem or a curriculum problem. It is a systemic failure of alignment between what schools teach, what they test, and what students actually retain. No proctoring software fixes that.
The conventional reform chorus will say: return to rigor, ban the AI, grade harder. But the UNAM’s experience suggests that the real issue is not the tool but the trust model. A proctored exam is a trust model from the 19th century. An online exam with no verification is a trust model from a fantasy. The realistic alternative is not to choose between them but to redesign assessment around what can be verified at scale—project-based portfolios, oral defenses, iterative writing with tracked drafts—methods that are harder to fake because they require sustained engagement.
The tradeoff is real. These methods are labor-intensive and expensive. They require faculty to do more than grade multiple-choice bubbles. But the consequence of not doing it is already visible: a university re-testing hundreds of students, a new degree whose graduates will be judged by follower counts, and a generation where half are left behind.
The decision that matters most is whether institutions will invest in assessment that is worth trusting, or continue to outsource that trust to technology that cannot deliver it.