7 Step Process for Deterring AI Misuse

1. Establish Clear AI Policies & Communication

• Draft or adapt a syllabus statement (e.g., from CITL’s Syllabus template) that:

  • Defines permissible vs. prohibited AI uses (e.g., grammar tools vs. generative drafting)

  • Requires students to disclose and cite any AI assistance (e.g., via an "AI appendix")

  • Explains the consequences of misuse

• Share policy in multiple venues—syllabus, Course Site announcements, first-day slides, "academic integrity" vignettes.

• Host an early-semester orientation or discussion forum on AI: capabilities, limitations, and ethics.

2. Rethink Assessment Design (AI-Resilient / AI-Organic)

• Process-Oriented, Scaffolded Assignments

  • Break major projects into milestones with deliverables: topic proposals, literature reviews, design sketches, code snippets, and interim reports.

  • Embed frequent low-stakes drafts and in-class work sessions with peer or instructor feedback.

  • Authentic, Contextualized Tasks

    • Center assignments on "the now": student backgrounds, local data sets, emergent engineering challenges, and institutional partnerships.

    • Require reflections on how AI was (or wasn’t) used, and critical analysis of AI outputs vs. human reasoning.

• Public & Collaborative Audiences

  • Build in peer-review workshops, online portfolios, symposium presentations, or real-world client feedback.

  • Encourage teamwork so students rely on human collaborators as well as AI.

3. Integrate AI Critically & Responsibly

Open

• Introduce "AI teammate" exercises: students perform a task manually, then repeat it with AI support, compare outcomes, and document trade-offs.

• Use CITL's "AI Sandbox" (setup requires a Ticket to the CITL Team), which allows instructors to see students' AI Chat History (within the sandbox)--teach prompts, effective uses.

• Teach students to evaluate AI suggestions for bias, accuracy, and relevance—fostering "AI critical" skill sets.

• Develop a small repository of vetted AI tools and usage guidelines tailored to engineering workflows.

4. Don't Count on Detection Technologies

• Many AI checkers deliver a concerning number of false positives, plus it has bias risks, privacy concerns, and arms-race dynamics with generators.

• If you use detection software, calibrate thresholds conservatively, communicate its fallibility, and always follow up flagged cases with human review, mindful of the impact you can have on students if you falsely accuse them of dishonesty.

• Lockdown browsers are available, but they create stressful conditions for students, harming classroom climate--and are defeatable by students who see them as intrusive.

• Due to open access to detection tools and AI text humanizers, there is currently no foolproof technology to detect AI misuse.

5. Faculty Development & Community Building

• Run workshops modeled on "Creating AI-Resilient Assignments": share best practices, sample rubrics, and co-design sessions for course-specific tasks.

• Maintain a living notebook (e.g., shared NotebookLM) of research, policy templates, assignment exemplars, and case studies.

• Establish peer-mentoring or learning communities for instructors to troubleshoot AI-related challenges mid-term.

6. Student Support & Well-Being

• Acknowledge stressors—time pressure, impostor feelings—that drive AI misuse. Provide resources: writing centers, counseling, and time-management seminars.

• Offer optional "AI skill labs" where students learn to use generative and analytic tools ethically and productively in engineering contexts.

7. Continuous Assessment & Iteration

• Collect mid-semester student feedback on AI policies and assignment workload.

• Track indicators of academic integrity incidents and correlate with changes in policy or assessment design.

• Iterate annually: refine syllabus language, adjust scaffolding cadence, update tool lists, and share outcomes across departments.