Student AI Use: Setting Expectations
Overview and Introduction: The WHAT and WHO
Student use of generative AI is no longer emerging — it is widespread. Nationally, student AI use skyrocketed from 22% to 85% between 2022 and 2025 [1], and at ASU, 88% of students reported using at least one AI platform [1]. Students report using AI primarily for research, summarizing, brainstorming, and drafting rather than wholesale outsourcing. At the same time, students are navigating uneven and sometimes contradictory expectations across their courses, which can leave them anxious and unsure whether they are crossing a line.
A 2025 multi-institutional survey of more than 16,000 students and faculty across six public universities found that
- ~52% regularly use AI-powered tools or applications in their studies
- ~51% would feel embarrassed if someone found out that they used AI for schoolwork
- ~33% agree that their professors encourage the use of AI in coursework [2]
Meanwhile, faculty stances on AI vary widely, and that variation is reasonable during a period of technological transition. The same survey found that 15% of faculty forbid AI use, 19% discourage it, 40% remain neutral or don’t address it, 18% encourage it, and 6% require it [2]. Notably, 62% report including explicit AI statements in all syllabi, while 22% include none [2].
These findings point to a critical insight: students are already using AI as part of their learning process, but they may not know what is acceptable in a particular course, assignment, or learning context.
This QRG helps faculty communicate clear expectations for student AI use so students understand when AI is permitted, limited, discouraged, or inappropriate, and why those boundaries support learning.
Implementation and Timing: The WHEN, WHERE, and HOW
Set AI expectations early, repeat them at key moments, and connect them to the learning goals of the course or task. A syllabus statement can establish the broad course approach, but students also benefit from reminders and examples when they begin specific assignments, projects, labs, exams, writing tasks, or collaborative work.
Aligning AI Expectations With Learning Goals
When drafting AI expectations, consider three guiding questions:
- What kind of thinking or skill should students demonstrate?
- Where could AI support that learning process?
- Where would AI replace, hide, or weaken the learning you need to assess?
AI expectations do not need to be long or legalistic. They need to be aligned with cognitive engagement and written in a student-friendly manner:
| Stance | Example Language | Reasoning Provided to Students |
| 🟥 AI may replace intended thinking | “Do not use generative AI to produce explanations or solutions.” | “This assesses your individual understanding of the course concept or skill; Seeing your specific reasoning allows me to give you meaningful feedback.” |
| 🟨 AI may support preparation but not substitute final work | “You may use AI to brainstorm or clarify ideas while drafting.” | “Your final submission must reflect your own analysis, and you should be ready to explain your process.” |
| 🟩 AI use aligns with the learning goal | “AI can be used to explore alternatives or debug code.” | “Engineers use these tools professionally to test and refine solutions. Include a brief note on how the AI influenced your thinking.” |
In each case, the expectation is paired with a reason. This reduces ambiguity, supports fairness, and makes the connection between tool use and observable thinking explicit.
Examples of AI Expectations in Engineering Coursework
Students benefit from seeing what appropriate AI use looks like in context, especially in courses where problem-solving, coding, writing, or design work occurs outside instructor visibility. For each task, describe how AI may support the work and where it should not replace student thinking. Expectations should be adapted to match course goals and learning objectives.
| Task | What appropriate AI Use could look like: |
| Timed quiz or exam (e.g., circuits fundamentals) | No AI use. This work assesses individual understanding of core concepts and problem setup. |
| Multi-step problem (e.g., beam loading, fluid flow, circuit analysis) | Set up the governing equations and assumptions independently. AI may be used to check algebra, visualize results, or suggest alternative solution paths. Be prepared to explain each step. |
| Developing code for a model/simulation | Outline the algorithm or pseudocode independently. AI may help implement functions or structure code. Document how AI suggestions were evaluated and modified. |
| Concept review (e.g., heat transfer modes, stress-strain relationships) | Use AI to generate practice problems, summaries, or alternative explanations. Cross-check with course materials and apply concepts in new contexts. |
| Reading technical material (e.g., textbook chapter, research paper) | Use AI to summarize sections or generate study questions. Verify accuracy and connect summaries to equations, diagrams, and course examples. |
| Interpretation of experimental/simulation results | Analyze trends and explain results independently. AI may help compare interpretations, identify possible errors, or suggest additional factors to consider. |
| Lab report (e.g., materials, fluids, thermodynamics) | Draft analysis and conclusions based on your data. AI may be used to support clarity, organization, or grammar feedback, but should not generate results or interpretations. |
In all cases, AI use should be acknowledged when relevant, outputs should be verified, and students should follow assignment-specific guidance for citing AI-generated content. (e.g., Purdue OWL guidance on citing generative AI).
Quick Faculty Checklist
Before publishing a course policy or assignment prompt, check whether students can answer:
- Is AI allowed, limited, or not allowed?
- What kinds of AI support are acceptable?
- What kinds of AI use would undermine the learning goal?
- Do students need to acknowledge, document, or cite AI use?
- What should students do if they are unsure?
Rationale and Research: The WHY
The Transparency in Learning and Teaching (TILT) framework emphasizes that students benefit when instructors clearly communicate an assignment’s purpose, task, and criteria for success [3], [4]. Research on TILT confirms that transparent communication of purpose, tasks, and criteria improves student performance and success rates — particularly for first-generation, low-income, and underrepresented students [3][4]. In AI-enabled learning environments, transparency should also include expectations for AI use and how it relates to the intended learning. Even one or two sentences connecting the AI expectation to the learning goal can reduce confusion and increase student trust.
For faculty, setting explicit expectations creates alignment between instructional goals and student behaviors. It reduces ambiguity in evaluating student work, supports academic integrity, and allows instructors to intentionally incorporate or limit AI in ways that reinforce the skills and thinking their courses are designed to develop.
Additional Resources and References
[1] W. Anderson, M. Angilletta, A. Hale, D. McNamara, E. Reilley, K. Rutherford, M. Sears, and J. VandenBrooks, “ASU AI Landscape: Patterns, Perceptions, Performance of Undergraduate Online and Campus Students,” presented at FOLC Fest 2026, Tempe, AZ, USA, Feb. 6, 2026. Available: https://docs.google.com/presentation/d/1fdkeiu9qS9mEo7K9fE3UwwbxC5vCGfDvg1sPBxWywJ4/
[2] D. Goldberg, J. Frazee, S. Hauze, and E. J. Sobo, “SDSU AI Student Survey Dashboard,” San Diego State University, 2024. [Online]. Available: https://aaai.sdsu.edu/initiatives/ai-survey. [Accessed: Mar. 13, 2026].
[3] M. Winkelmes, A. Bernacki, J. Butler, M. Zochowski, J. Golanics, and K. H. Weavil, “A teaching intervention that increases underserved college students’ success,” Peer Review, Association of American Colleges & Universities (AAC&U), vol. 18, no. 1/2, pp. 31–36, 2016.
[4] M. Winkelmes, “Transparency in teaching: Faculty share data and improve students’ learning,” TILT Higher Ed, 2025.