More than half of U.S. college students now use AI in their coursework at least weekly, according to an April Lumina Foundation-Gallup study. The numbers are forcing colleges to move past outright bans and toward structured policies that teach students how to use the technology without outsourcing their thinking.
Of students who use AI, 54% frequently use it to edit or improve their writing, 49% for generating topics and ideas, and 36% to write entire papers. Pittsburgh-area schools still prohibit students from having AI fully write a paper. But they are carving out room for limited, disclosed use.
"Initially, AI was feared. How do we prevent the use of AI from an academic integrity standpoint?" said Brian Danielson, Slippery Rock University's senior director of teaching, learning and academic operations. "That's starting to shift now. Faculty are doing more and more to build AI in teaching. We're taking a proactive approach in how AI can be used in the writing process."
Professors set the boundaries, assignment by assignment
At Slippery Rock, Robert Morris University, and other area schools, the decision about AI use rests with individual professors. Robert Morris has developed an AI policy, a teaching handbook, and training sessions to help faculty decide where the line falls. Jon Radermacher, vice provost for academic administration, said the Moon Township school promotes academic freedom while requiring transparency: "If students use it, they have to disclose how they used it."
Danielson said AI can handle brainstorming, searching for studies, checking grammar, or giving feedback on a draft. But professors must be explicit about what is off-limits - and explain why. "Most importantly, they explain why it's OK or not OK, and it should be very explicit about how AI is used for this specific assignment," he said.
Camille Dempsey, executive director of Pennsylvania Western University's Center for Artificial Intelligence and Emerging Technologies, said using AI ethically often takes more time than the traditional approach. "The better way to teach writing practices using AI is to jump in with students - explore tools, what it generates, what's acceptable and what's not," she said. "The best practice is when faculty members work with students on AI for Writers literacy."
What AI cannot do
Edward Karshner, English and Media Arts Department head at Robert Morris, compared current anxieties to Socrates' hesitancy about written texts. The ancient Greek philosopher feared books would weaken memory and create false wisdom. Karshner said faculty should design assignments around what AI lacks: "AI can't discern. It doesn't have a motive and it doesn't take into account an audience. When AI generates texts, it's incomplete. Students have to know how to make sense, and how to make use of that."
Melissa Hrico, an English and creative writing professor at the Community College of Allegheny County, counsels students against using AI for idea generation or brainstorming. "Humans come up with stronger, more creative ideas," she said. She has students compare feedback from a human tutor and an AI tutor. "You don't want to have Gemini or ChatGPT write anything for you, then you'll lose your voice as a writer. Students understand it is a useful tool, and it does work well, but it's up to human discretion if they want to accept advice or keep it as-is."
Students who submit AI-written papers fail the assignment. Danielson said the real cost is to the student's own development: "The student robs themselves of the opportunity to learn. It's removing the student's ability to practice critical thinking and higher-order thinking skills."
Redesigning assignments for an AI era
If a student can complete an assignment by pasting a prompt into ChatGPT, the assignment itself needs rethinking, Hrico said. "Hiding under a rock doesn't help. AI is here to stay, and we don't want to be ignorant. It's like how the Internet, and Zoom classes, changed things."
Robert Morris supports faculty in building "AI-resistant" lessons - using newer texts not widely available in public databases, for example. Slippery Rock pushes faculty to craft writing prompts that demand problem-solving and creativity, skills AI cannot replicate. "By reenergizing writing prompts, you encourage the student to think more critically and not use AI," Danielson said.
Some professors now assign comparison tasks: students write a paper, then critique an AI-generated version of the same prompt. Style differences become a teaching tool. The approach is spreading across disciplines as AI for Education moves from a compliance problem to a curricular one.
Why this matters for writers
The classroom experiments unfolding now will shape what entry-level writers bring to the workforce. Employers will inherit graduates who learned to use AI as a feedback tool, a grammar checker, or a research assistant - but also graduates who never developed the ability to generate original ideas under their own power. The distinction between augmenting writing and replacing it is being set right now, syllabus by syllabus. Writers who understand where that line falls, and can articulate their own process for staying on the right side of it, will have an edge in a market where the tool is universal but judgment remains scarce.
Your membership also unlocks:
