To be fair, Nataliya Kosmyna and colleagues actively tried to correct misinterpretations. On LinkedIn and their project website, they repeatedly emphasized that the study did not show LLMs make people stupid, and that their findings were preliminary. They clearly listed limitations: small, local sample; essay-writing only; no peer review yet. The team also promoted the preprint through short interviews and explainer videos (CNN, Time, CBS). While these were meant to clarify, they arguably magnified the hype by repeating ambiguous terms and reinforcing the most dramatic interpretations.

But here’s the uncomfortable question: If a study is so easily misinterpreted, isn’t that partly a failure of communication? A 200-page preprint is nearly impossible for most readers to parse, and terms like cognitive debt, while catchy, were never clearly defined. To make matters worse, the framing oscillated between warning of a “possible decrease in learning skills” and describing “nuanced connectivity differences,” sending readers conflicting signals and leaving the door wide open for sensationalist headlines.

The methodology itself raised questions. Key details, such as the exact prompts given to the LLM group (by students, not the one on page 46 used by the authors), or how instructions framed to participants were vague. Without these, interpretation becomes tricky. For example, did participants think they should copy-paste outputs or use them creatively? The absence of clarity opens the door to misinterpretation and weakens the validity of comparisons across groups. In short, poor communication choices hurt both the audience (who walked away misinformed) and the authors (whose work was sensationalized and criticized).

The lead author repeatedly justified releasing the study early, before peer review, because of the “intense speed” of developments in AI. Yet this points to a deeper tension: are we publishing for understanding, or for visibility? “Publish or perish” pressures, combined with the allure of media attention, risk fueling hype while shifting responsibility onto “the speed of the field.”

In one interview, Kosmyna mentioned that while many educators wrote to them, no AI companies did, and urged the industry to “be on the correct side of history.” But what does that even mean, when the study itself was rushed out and amplified its own negative hype? Can companies be blamed for not endorsing a paper that was released in a preprint with unresolved flaws?

The authors have also pointed to the long peer-review process as a reason for going public early. It is true that peer review can take months, even years. But that very slowness generally contributes to rigor. Careful review helps filter out unclear methods and prevent misleading claims from spreading unchecked.

This lack of accountability is visible across interviews and posts. Rather than acknowledging their own role in choosing ambiguous terms, selective framing, and early publicity, the authors often portrayed the misinformation as entirely the audience’s fault. But communication is part of science, and responsibility cannot be outsourced.

Another striking point: most preprints never receive this level of media coverage. The unusual buzz here likely stemmed from the MIT brand, dramatic framing, and the provocative title. This shows how institutional prestige and catchy wording can catapult preliminary findings into viral headlines.