AI in the classroom: why lowering standards is the worst answer to students who won't read

By Zendoric · September 15, 2026.

Audrey Campbell, a senior at American University's School of International Service, published an opinion column in the student newspaper The Eagle that portrays, from inside the classroom, a phenomenon that global statistics only hint at: universities are lowering academic demands to accommodate the massive use of generative AI among their students, instead of requiring them to use it without giving up reading or thinking for themselves.

The account begins with her exchange semester in Berlin. As she tells it, one professor allowed the use of AI on the final essay, merely remarking that he would find it boring to read; another handed out a printed ten-page article and then also offered the digital version "in case they wanted to run it through AI." That same semester, Campbell recounts, an instructor reduced the required reading list to "read, skim or summarize at least one" — a gesture she frames within the trend of professors cutting readings because students refuse to do them. The anecdote that sums up her unease: after turning in her own definition of "energy policy," a professor doubted it was her work. "You didn't write that, did you?" he asked her, according to the column.

To argue that the problem goes beyond her personal experience, Campbell cites third-party data: according to UNESCO, in 2026 only half of the world's students reach the expected reading level by the end of primary school — a gap she herself acknowledges predates generative AI and has broader causes. To that she adds statements from teachers in England in early 2026 about students with basic deficiencies in writing and problem-solving, and a late-2025 study — 580 participants, with no further methodological detail in the article — linking heavy AI use with lower critical thinking; 74% of those participants said they used AI mainly for work and study.

It is worth placing this precisely: it is an opinion column signed by a student, not academic research, and the newspaper itself flags as much at the top of the text. Her supporting data are heterogeneous — a global literacy statistic, a modestly sized study, scattered statements from teachers — and do not amount to causal proof that AI is "atrophying" an entire generation. But the pattern she describes — professors who give in to student resistance and end up asking for less, not more — fits what has been observed in higher education since language models became capable of summarizing, drafting and arguing with ease.

Here is the tension worth flagging. At Zendoric we have argued that the education that wins with AI is the one that trains teachers capable of orchestrating it as a tool for augmented learning, not the one that gives up on demanding effort from students. This testimony shows the dark side of that orchestration when it is done badly: if "use AI if you want" replaces "read this and then we'll discuss it," the tool does not free up cognitive time to think better — it eliminates it outright. Long-form reading, unscaffolded writing and critical thinking are not curricular ornaments: they are the mental infrastructure on which any sophisticated use of AI is later built, including the kind that underpins the promise of abundance we are betting on over the long term.

That is why we believe the right answer is neither to ban AI in the classroom — broadly speaking, that route already failed when many schools blocked access to ChatGPT in 2023 without really curbing its use — nor to surrender and cut readings because nobody does them. It is to redesign assessment so that a student's own effort remains irreplaceable: oral exams, verifiable in-class work, documented processes, assignments that AI cannot solve on its own without it showing. In the short term that demands more work from teachers, not less — exactly the opposite of the temptation Campbell portrays. In the long term, the generation that learns to think with AI rather than letting it think for them will be the one that best exploits the abundance this technology promises: whoever knows how to formulate the right question will still have the advantage, even when the answer is one prompt away.