Industry Corner: AI in higher education - help or harm?
AI use in higher education has moved from side project to core system. Colleges are reworking curricula, experimenting in the classroom, and rethinking assessment - while trying to keep ethics and critical thinking intact.
The balance is the work. Faculty need to prepare students for an AI-heavy job market and teach them how to think clearly about the limits, risks, and appropriate use of these tools.
Where AI use stands now
Student adoption is widespread: 90% of graduate and undergraduate students use AI, and 73% increased their use in the past year, per the 2025 AI in Education Trends Report by Copyleaks.
Many institutions are moving in step. "AI is unquestionably the biggest driver of change," said Isabelle Bajeux-Besnainou, dean of the Tepper School of Business at Carnegie Mellon University. "Across the board - from operations to instruction to outreach - AI has become deeply embedded in everything we do."
The real tension: skills, ethics, and incentives
Students want an edge and often test the limits of acceptable use. As one dean shared, prospective students are asking how programs will prepare them for jobs in 2040 and beyond - not just the next hiring cycle.
Pressure is real. New college grads faced higher unemployment than the broader population from May 2024 to May 2025 (about 6.6% vs. ~4%), with some reports putting recent grads near 10%, while tech layoffs tied to AI shifts haven't helped. Employers are raising the bar on how graduates use AI, not just whether they can use it.
Policy clarity beats guesswork
Surveys show many students don't know what is allowed and what crosses the line. Growth in AI tools is outpacing course policies in plenty of programs.
Some schools are catching up with program-by-program rules. As Nossi College of Art & Design's leadership explained, the key is knowing where AI adds value and where it creates risk. In coding courses, use may be broader; in creative classes, it may be limited to research.
That approach pairs skill with judgment. As Debra Schwinn, president of Palm Beach Atlantic University, put it: students should think for themselves and use AI with care.
A practical playbook for academic leaders
- Set course-level AI policies: Define allowed, restricted, and prohibited uses with examples. Post in the syllabus and LMS.
- Teach the tool, teach the limits: Cover error patterns, bias, hallucinations, citations, and data privacy. Make students show how they verified outputs.
- Assess the process, not just the product: Require drafts, version history, prompts used, and oral checks or live coding/writing sprints.
- Build AI literacy across majors: Short modules in first-year seminars and capstones. Calibrate depth by discipline.
- Create an ethics decision tree: What's acceptable collaboration with AI? What needs disclosure? Where is it off-limits?
- Adopt disclosure norms: Simple statements: tool used, purpose, prompt summary, what was kept or corrected.
- Invest in faculty upskilling: Micro-trainings, office hours, a shared prompt library, and course exemplars.
- Protect academic integrity: Use mix of policy, assessment design, and conversation. Tools help, but the culture does the heavy lifting.
- Center privacy and IP: Clarify data handling, institutional accounts, and student consent before tool use.
- Keep humans in the loop: Use AI to personalize and speed feedback, while reserving judgment, grading standards, and mentorship for faculty.
Course-use guidelines (quick reference)
- Coding and data: Allowed with disclosure; require tests, comments, and error analysis. Spot-check with live sessions.
- Writing-heavy courses: AI for brainstorming and structure with disclosure; final voice and citations must be the student's.
- Design and media: Permit research and mood boards; restrict full generative outputs unless the course teaches those tools explicitly.
- Quantitative exams: Use proctored or in-person assessments. For take-home, require work shown and oral defense.
90-day rollout plan
- Days 1-30: Draft institutional AI principles, legal guardrails, and a standard syllabus clause. Identify pilot courses in 3-4 departments.
- Days 31-60: Train pilot faculty; publish course policies; redesign one assessment per course to include process evidence.
- Days 61-90: Run pilots; collect student/faculty feedback; report outcomes; adjust policy and scale to more sections.
What to tell students
- AI is allowed where stated, with disclosure. Use smartly, show your reasoning, and cite your sources.
- Your edge is judgment. Employers want people who can question outputs, spot risks, and make sound calls.
- Build durable skills: problem framing, data literacy, prompt craft, model verification, and clear writing.
- Own your work: AI can assist; it should not replace your thinking.
Keep the human at the center
Carnegie Mellon's view is clear: as systems get smarter, human judgment must get stronger. "We want to prepare learners to critically assess, cross-check, and apply AI - not just use it mindlessly…. Education should inspire human agency, not erode it," said Bajeux-Besnainou.
Resources
- 2025 AI in Education Trends Report (Copyleaks)
- Recent grads face tougher hiring conditions (WSJ)
- Curated AI course paths by job role (Complete AI Training)
Your membership also unlocks:
