/* Change the "Read More" text on blog posts */

A student came to my office needing help with a paper. When I asked to see her draft, she turned her laptop toward me: a blank page, a blinking cursor.

“I just don’t know how to start,” she said, and started to cry.

What was shocking were the pages of handwritten notes she had. She clearly knew the material. She simply could not find a way to start.

I asked whether she had tried an artificial-intelligence tool. She replied, “Professors scare me…they say it’s cheating.”

I gave her permission to photograph her notes and then prompt ChatGPT to arrange the ideas into a workflow. Seeing the scaffold that the AI provided, her panic subsided. She began typing.

“Okay,” she said. “I know what to do now.”

She hadn’t been looking for a shortcut; she simply needed a foothold. Embarrassingly, in all those years, it had not occurred to me to build one; yet that is my job as an educator.

As professors craft their fall syllabi, course policies are hardening. The reaction is understandable. The timing, however, is grotesque.

Just as the federal backstop that made colleges take accommodation seriously has weakened, colleges are tightening the rules on the only help some students have.

Twenty-one percent of American undergraduates report a disability, many of them neurodivergent. Among students who have a disability while in college, only 37 percent inform their school. The rest may be getting nothing. Many were never diagnosed, can’t afford the testing that produces documentation, fear the stigma, or won’t trade sensitive medical information for uncertain help.

Another of my students, who identifies as autistic, had no idea what my margin comment “develop a critical response” meant. I’d been scribbling that phrase on papers for a decade. It assumes an academic vocabulary instructors never explain. On his own, he asked an AI to translate it into steps he could understand.

The machine did not do his thinking. It did mine. It gave the explanation I’d been too rushed, or too oblivious, to offer.

Until now, educators had a plausible excuse to be relatively oblivious to students facing learning barriers: individualized support is expensive. It takes tutors, specialists, learning centers, appointments in narrow windows. But notice what the excuse concedes. We always knew what these students needed. We decided we couldn’t afford it for all of them — so we built an obstacle course of bureaucracy and rationed it to those who made it through.

Recently, help got cheap. A student can now ask AI to break an overwhelming task into steps or put a chaos of notes in order: privately, at two in the morning, without a diagnosis, without a form, without disclosing anything.

Our first response was to forbid it.

The shortcutting students and the scaffolding ones use the same software, yet they are doing very different things with it. A rule that treats those cases the same is a failing education policy, where indifference masquerades as rigor.

 The image displays a portrait of a smiling man with a goatee and short brown hair and he is wearing a patterned tie and light blue dress shirt.

Matthew Brophy

Some professors will rightly forbid AI where the machine performs the very skill being taught. But none of this discharges the duty of care colleges owe these students: a way to access their education — an equitable opportunity to learn.

We must stop fixating on whether a student used AI. We must ask what she used it for. Did it carry her past the learning? Or, like the student crying in my office, did it finally provide her with an equal opportunity to learn?

Matthew Brophy is an associate professor of philosophy at High Point University, where he teaches ethics and technology. He is the author of “The Case for AI as Accommodation” in Inside Higher Ed. His writing has also appeared in The Chronicle of Higher Education and EDUCAUSE Review.