The rapid improvement of generative AI tools has led many of my peers to proclaim that higher education as we know it has come to a crashing and shocking end. I agree.
In my large-enrollment general education course at the University of Florida, I can no longer assign an essay asking students to state their views on genetic engineering and assume the responses I receive are written by humans. So the critical question we must ask as academics is, “What do we do now?”
Rather than try to create assignments that AI cannot tackle, I propose we develop assignments that embrace AI text generation.
We don’t want to ignore the 54 percent of students who use AI at least weekly in their course assignments, according to the Digital Education Council. We don’t want to ban AI. And even when we, as educators, try to trick AI tools, newer versions of ChatGPT just come along to thwart that strategy.
With this in mind, I modified the final assignment in my course to require that students submit an entirely AI-generated first draft, which they then modified to reflect their own perspectives. In the first couple of semesters using this strategy, students color-coded the sources of text to mark which parts were human-generated and which were AI-generated. This strategy allowed students to use and reflect on how they would utilize AI in the future.
Tracking of text origin was further streamlined using the recently released “Authorship” tool from Grammarly, which accurately attributes text as “typed by a human” or “copied from a source/AI-generated.” Advancements in technology have upended the careful development of assessments in higher education before and will continue to in the future, even if AI appears to be an all-encompassing, do-everything tool.
For those of us born in the 1970s, we remember a time before the ever-present calculator. Math teachers could assign long-division problems without worrying that students who came up with the correct answer did not understand the methods required to generate the answer.
More recently, language translation, a key learning tool in language acquisition, was upended over a few days in 2016 by the release of a new version of Google Translate. The rapid improvement in Google Translate bears similar parallels to how ChatGPT 3.5 burst into the consciousness of a large portion of the population in November 2022. In both cases, educators eventually embraced and used these new tools to improve student learning outcomes.
While requiring a GenAI first draft of an assignment is not a model that will work in all situations, “showing the work” and student reflection can play key roles in student assessment. My twin high school seniors possess graphing calculators that are more powerful than the computer on which I wrote my dissertation. So I have observed firsthand how educators have modified assessments to adjust for such changes, emphasizing the processes needed to answer the assignment more than the final answer.
Language teachers have pivoted to incorporate student reflections on why one word was chosen over another, for example. In my course, Part B of the final assignment requires students to reflect on how well (or poorly) the initial AI draft reflected their views on the assigned topic.
I acknowledge that students with access to AI during the reflection portion of assignments could use the tool to show how they produced the “work.” Tools that track AI usage, like the “Authorship” tool, hold promise for providing both instructors and students with information on where and how much AI text was used in an assignment.
The capability of AI to generate text (and images) will keep advancing, becoming increasingly integrated into the daily lives of both us and our students. In our professional lives, it will be capable of responding to the most imaginative essay prompts educators can design.
By shifting the focus of assignments from pure content creation to critical engagement, analysis and editing, we will teach our students how to think creatively, collaborate and communicate their ideas effectively and responsibly. These are the same skills they need to master to successfully work in teams and communicate efficiently in their future careers.
Brian Harfe, Ph.D., is a professor in the College of Medicine and associate provost at the University of Florida. He runs 14 international exchange programs and a study abroad program while teaching about 450 students each semester.
