Center for Online Learning

Popular Opinion: We just can’t give first-graders calculators, right? Likewise, higher-ed students need to learn the fundamentals of math and science, free from generative AI or else they will become overly reliant and have a weak conceptual foundation to build upon, and what if they’re ever put in a situation where AI is not available? They’ll be helpless!

Raise your hand if you can calculate the sine of 42.2 degrees without a calculator or reference. 

Pretty rare if you can do this with precision. Hipparchus developed tables to look up the value of trig functions. A couple thousand years later these were functionally replaced by electronic calculators, and today’s practitioners no longer have physical copies of the tables. Likewise, many beginner computer programmers are now starting out on feature-rich IDEs (integrated development environments) rather than more primitive editors.

How do we decide if a technological shortcut abstracts essential methodologies learners need to grow?

AI Will Tempt Students to Avoid Learning

If a student has free rein to use AI during a “traditional” general physics exam, that student is likely to get a good grade. And we’ll have no idea if they learned a single thing! STEM faculty are at a crossroads where they need to make decisions on what students need to learn versus what students need to learn to accomplish. 

This is tricky.

For now, however, let’s assume that we still need students to perform the same learning objectives we’ve been using. This necessitates AI-free moments for assessment, while also allowing students to benefit from AI’s usefulness as a learning aid.

This post offers a practical, flexible model for integrating AI into a lower or mid-level STEM course. The goal isn’t to overhaul your course overnight, but to help you make thoughtful adjustments that acknowledge the presence of generative AI while still supporting meaningful learning, academic integrity, and student growth. Each suggestion is designed to be immediately applicable, adaptable to your teaching style, and grounded in real classroom experience.

For a similar post with a focus on humanities courses, take a look at:
Michelangelo Had a Chisel – Your Students Have ChatGPT

What an AI-Aware shift could look like

Encourage AI Use for Homework

Encourage your students to use AI to assist them on their homework (Ajani, 2025). You might even consider providing an AI tutor customized to your course (Mowreader, 2025). Some students will be tempted to abuse this and transfer AI output mindlessly into their work. The counter to this is a clear warning that doing so will cause them to be exposed later. The goal is to use AI to empower students to overcome gaps in their own understanding. This will require some demonstration and coaching from you.

Quizzes

Quizzing can be done in a number of ways. For example, you might still do them in class, or they can be random-draw video-response quizzes where the student must explain the steps and rationale for a problem. Use these to validate that the students’ understanding of the homework is genuine and personally address discrepancies between homework and quiz performance prior to a disastrous exam performance.

Pre-Tests (Studying in Disguise)

There’s students that do well, and there’s students that would do better if they studied more. We can, in a sense, trick them into studying more through pre-tests (Roediger, 2006).

Here’s one approach: Give them a random draw pre-test prior to the real exam. Make it slightly more challenging than the actual exam (Bjork, 2011). Suggest they use AI to learn how to overcome deficiencies; however, once again caution them that if they cannot perform on this exam without AI, they will fail the real exam. 

Warn them that if their grade on the pre-test is dramatically higher than their grade on the real exam, that you will have a conversation with them about why that is. 

In-Person or Proctored Exams

Here’s where we see if the formative prep and warnings were successful. In most cases these will be in-class exams. You can consider using Respondus LockDown Browser paired with collecting hand-written steps on paper. Or you can use Gradescope to create testing sheets that can, with surprising ease, be ingested into Canvas! This captures both discrete answers and student handwriting. Dr. Trisha Renner from the UD Physical Therapy department has found massive time-saving efficiencies in her exam grading practices thanks to Gradecope. 

You may also consider coupling your exams with Interview Exams. These give you an opportunity to personally connect with your students and measure what they truly know (Qi, 2021)! 

Semester-Long Creative Project

A Purdue-Gallup study of 30,000 college graduates identified semester-long projects among the "Big 6 Experiences" that best predict student satisfaction and success with life and careers (Gallup 2015). What’s something experiential, inspiring, or personal that can connect the student to the content? Build a project around that.

Encourage the students to use AI in the same way a practitioner would out in the field. Lastly, requiring a public presentation of their project heightens the stakes and should push them to put in the effort and add value to their university journey.

Adjust Expectations for Upper-Level Courses

These recommendations are primarily intended for lower and mid-level STEM courses. Generally, the level of intrinsic motivation of the students paired with the greater complexity of upper-level courses reduces AI exploitation concerns. The nature of advanced topics inherently demands a deeper level of engagement and critical thinking. Meaning AI is a great help but can’t create course deliverables entirely.

Final Thoughts

If this is a bit overwhelming just start with the one or two assignments that were the biggest problem last time you taught the class. The most important thing you can do is set expectations and carefully design your learning activities to encourage students to thrive within parameters optimized for your learning goals. 

Hipparchus used an astrolabe to predict the future positions of celestial objects. Unfortunately, no such device exists to predict where education will evolve from here.

But you’re not alone! If you need any help at all, reach out to onlinelearning@udayton.edu. We’re here to support you—whether you want to brainstorm one assignment or rethink your whole course.

— Written by Paul Dagnall, Director of Online Learning at the University of Dayton 

References

Ajani, S. (2025). Leveraging Artificial Intelligence as a Learning Tool in Higher Education. International Journal of Educational Research, 7(1), 1–17.

Bjork, Robert A., and Elizabeth L. Bjork. “Creating Desirable Difficulties to Enhance Learning.” In Psychology and the Real World: Essays Illustrating Fundamental Contributions to Society, edited by Morton Ann Gernsbacher, Richard W. Pew, Leaetta M. Hough, and James R. Pomerantz, 56–65. New York: Worth Publishers, 2011.

Gallup. “Big Six College Experiences Linked to Life Preparedness.” Gallup News, April 8, 2015. https://news.gallup.com/poll/182306/big-six-college-experiences-linked-life-preparedness.aspx.

Mowreader, A. (2025, January 22). Survey: College students enjoy using generative AI tutor. Inside Higher Ed.

Qi, Huihui, Shweta Singh, Minju Kim, and Wendy W. K. Mok. "Oral Exams: A More Meaningful Assessment of Students' Understanding." Journal of Statistics and Data Science Education (2021).

Roediger, Henry L., and Jeffrey D. Karpicke. “Test-Enhanced Learning: Taking Memory Tests Improves Long-Term Retention.” Psychological Science 17, no. 3 (2006): 249–55.