Before the start of the 2025-26 academic year, the Studio for Teaching and Learning Innovation at the College of William and Mary held a workshop on adopting Artificial Intelligence policies for courses. As AI becomes more advanced and applicable, STEM professors in particular have begun to integrate AI into their courses.
At the College, STLI provides generative AI resources for professors. Some programs that STLI has already approved include Microsoft Copilot, Adobe Firefly and Blackboard Learn AI Design Assistant. Other AI resources, such as ALEKS, are required for certain STEM courses, like Chemistry 103: General Chemistry I.
CHEM 103 teaching assistant Samantha Orr ’25 highlights the benefits of this program, where Gradescope autogrades multiple-choice and short-answer questions.
“It saves me from having to grade 200 problem sets 12 times a semester, which is really convenient for me,” Orr said.
While generative AI provides valuable homework resources for these classes, there are also concerns, such as environmental impacts and accuracy.
“I feel like sometimes it misses the mark, like relating to professor-specific questions,” Orr said. “A professor may explain it one way, but ALEKS explains it a different way, and then it doesn’t really match.”
Some syllabi state that this program often makes mistakes, which may impede its ability to take full responsibility for grading.
“I don’t think it should ever take the place of human touch,” Orr said. “I think it can speed up the process, like with grading, but I still think you should always have a person looking back over it and double-checking.”
Some computer science students find that generative AI decreases the workload at the expense of their learning.
Computer science student Chloe Toda ’29 explains how students view utilizing AI for assignments.
“It could be useful for other projects in the future if we actually knew what we were doing as opposed to just having the code doing it for us,” Toda said.
However, there are some benefits to using AI for coding.
“Sometimes it’s helpful to have AI generate certain test cases that you might not think of that will actually break your code,” Toda said.
Integration exists at varying degrees. Chemistry, for example, uses AI for the majority of its homework model. STEM-related COLL classes, on the other hand, have a different relationship with these materials, discouraging the use of AI for writing.
Assistant teaching professor Kelly Hallinger, who teaches a biology COLL 150 course, stated that the College is providing resources to all of these professors regardless of the course model.
“I attended a workshop run by the Studio for Teaching and Learning Innovation before the semester started, and one of the topics that was covered at that workshop was adopting an AI policy for your class,” Hallinger said.
STLI is just one example of how AI policies are becoming more commonplace in STEM courses. This year, it became mandatory for syllabi to include an AI policy. Some people view this as a negative shift from other traditional sources of insight and education.
“Take it with a grain of salt,” Toda said. “Go to real people for help instead of relying on AI all the time.”
There is a concern that students may overutilize AI.
“If we don’t teach students its limitations and we don’t teach them how to use it appropriately, then they’re going to use it anyway, but without oversight and guidance,” Hallinger said.
Even though AI is being used in this context for learning, it is still a new tool, and some argue that traditional materials should not yet be abandoned.
“I think that when you’re learning foundational skills, it’s important that you go through all of that struggle of the process yourself,” Hallinger said.
