Search engines did something similar to research. They did not abolish the need to know. They changed the burden from finding a needle in a library stack to judging whether the needle was real, relevant, and sharp. Columbia psychologist Betsy Sparrow and colleagues showed that internet search changed memory habits. People became more likely to remember where information could be found rather than the information itself.⁹ ¹⁰ That is not the end of intelligence. It is a change in the work intelligence has to do.
AI is another such shift, and probably a larger one. It can summarize, draft, translate, code, brainstorm, imitate, fabricate, and flatter. It can be a tutor, assistant, parrot, con artist, or mirror. That is why the old classroom model of “turn in the paper and I will infer the learning” is breaking down.
But turning school into a surveillance state is not education. It teaches compliance, fear, and concealment.
The Los Angeles Times reported that UCLA students described anti-AI exam rules involving mirrors, visible hands, and body-position monitoring. One student said, “It just felt so degrading.” Another said she would deliberately make her writing look more careless and unprofessional to avoid suspicion. A Los Angeles attorney representing students in college disciplinary cases said AI accusations now make up roughly 35 percent of her firm’s education caseload.¹
That is a remarkable sentence. Students are not merely learning how to write, think, and research. Some are learning how to look innocent to a detector, a professor, or a disciplinary board.
Vanderbilt saw enough risk in this system that it disabled Turnitin’s AI detector in 2023. Its explanation was careful, but the message was clear: a false accusation can be serious, and detection tools are not a substitute for judgment.⁶
This is where transparency can become its opposite. We say we want disclosure, but if disclosure means stigma, people will not become more honest. They will become more skilled at hiding.
The workplace version is already appearing. Some employees fear that admitting AI use makes them look lazy or replaceable. Some companies have unclear or restrictive policies, while workers quietly use AI anyway.⁴ ⁵ That is the predictable result of shame-based rules. They do not eliminate the behavior. They move it underground.
The better model comes from the other direction. Business Insider reported that SharkNinja paused regular operations for a four-day companywide AI hackathon. CEO Mark Barrocas said, “AI is the great equalizer,” and added, “Our job is to not leave anyone behind.”³
That is the right instinct. Do not leave people behind. Teach the tool. Teach the limits. Teach verification. Teach the difference between a draft and a decision, between assistance and authorship, between convenience and truth.