Each semester, Associate Professor of Biology Eric Luth sends a letter to students detailing why he makes distinctions in how AI is used in his classes. He urges them to value the power of communication via the written word, as well as the process necessary to improve that skill.
“I was trying to put myself in their position,” says Luth, about how he wrote the letter. “If I’m a busy, stressed-out college student and there is a technology that can write my papers for me, why should I waste my time on this? As a professor, I can’t just tell them to ignore the instant solution because I want them to do it the hard way; I need to tell them why, and tell myself why it’s worth it to continue to teach them.”
For this approach, Luth received a 2026 Dean’s Award for Excellence and Innovation in Teaching and Learning. He was commended for offering practical guidance in the use of AI, while making clear distinctions between AI that can support a student’s work and that which substitutes a student’s efforts to develop and learn.
The Push and Pull of New Technology
From a faculty member standpoint, detecting AI-written assignments, Luth says, is a losing battle. “I want to remove the incentives for students to use AI [to write their papers], and get them to see the value of writing things themselves.”
To do this, especially in his courses that are designated as “writing intensive,” Luth aims to emphasize the importance of the writing process.
“We start with the initial pitching of an idea, then fleshing out a full topic, before outlining and drafting a paper,” says Luth. “I want to guide students through the process of writing a paper, going through multiple drafts as opposed to one high-stakes deadline. This will help shape their writing, allowing them to see improvements over time.”
Aside from scientific insight, improvements to written communication are part of the goal of the project.
“You don’t need to be a professional writer for your career in order to benefit from writing well,” says Luth. “Writing and thinking are related; they inform each other and support each other. To have a mess of thoughts in your head and be able to order and structure them in a way that allows you to clearly communicate with others, to share your perspectives and illuminate your own thoughts, that will make you a better thinker and a better writer.”
The Use of AI in Education
While he has encountered a healthy amount of skepticism among his students about the use of AI, he feels strongly about the importance of discussing the temptation to use it.
“The more we build into our courses time to talk about this, the more we can discuss both the value and the risks of using AI,” says Luth.
It’s also important to make clear the distinction between generative AI, like ChatGPT and Gemini, and other types of machine learning platforms that can be trained to look for patterns and mine data sets.
“Machine learning can be incredibly useful for research. It can do things that we can’t do in such a time-efficient way,” says Luth. “My concern, particularly for the writing course, is around generative AI and the educational impact of its use, to say nothing of the environmental and ethical concerns around the technology, which are also considerable.”
As for how Luth’s students in “Animal Physiology” (BIOL 222) are permitted to use AI, it is useful for finding scientific articles to be used for data analysis and a class presentation.
“Finding an article at the appropriate level and topic can take up a lot of time,” Luth says. “I want them to spend the bulk of their time on the analysis and presentation.”
To that end, he said he includes instructions in the syllabus for creating a specific prompt to instruct AI to “do some of that digging that is necessary, but not the focus of the assignment, not where the true value is.” The prompt is engineered to generate tables and links to articles, but not to provide its own analysis.
For a semester-long grant proposal assignment in “Neurobiology” (BIOL 334), students use ConsensusAI, a specialized AI platform that only looks at published scientific papers, to create a brief literature review on their topic of interest. One output is a table, which states a claim and includes a list of papers that support that claim.
“Generative AI can write a grant review in a couple of minutes in a way that is difficult for me to police and detect,” he admits. “I knew that the temptation to use AI would be strong, so I chose to operate under the assumption that students would give in to the temptation, even if I told them not to. I redesigned the course to mitigate this, asking everyone to use AI and produce more than I would have otherwise, to level the playing field across the class.”
Students take a subset of those claims generated by ConsensusAI to validate themselves, finding the papers and statements in the text that support or don't support the claim. This approach gives students an “outlet” to use AI on one part of the project, so they can focus on their own work where it really matters to Luth.
“They can also use AI to understand experimental procedures that are unfamiliar, or analysis techniques that the authors used to analyze data,” Luth notes. “Neurobiology is so broad, there are procedures and technologies that students may read about that they haven’t learned in class. This allows them to supplement their knowledge, to learn more about alternative methods of analysis.”
The project includes a section on potential pitfalls of particular research methods, as well as potential alternatives. ChatGPT and other common AI tools can assist students in providing these alternatives. “There are limitations of each method or approach to an experiment,” says Luth. “If one technique won’t show what they want to show, they can offer multiple approaches to get the answer they are aiming for, even if we haven’t learned about them in class. It takes initiative to investigate unfamiliar methods and learn about options, but students will need to apply them appropriately on their own.”
Reflecting on the Benefits and Drawbacks of AI
Luth also polls his students about how they use AI, encouraging them to reflect on the process.
“Some were happy to have found a tool in ConsensusAI that they could see using in graduate school to identify useful articles,” he says. “Others are surprised to see how often the tool would overstate or overgeneralize evidence for a claim, or misapply data from primary sources. It’s good to see them analyzing that.” He is considering offering a future option that permits students to avoid using AI altogether, if that is their preference.
Luth encourages honesty in his post-assignment surveys, which the students fill out anonymously. “I ask how they used AI in the assignment, including options for things that I expressly prohibited in the instructions. Students were forthcoming about how they used the technology,” he says.
“Overall, they appreciate that I’m not putting my foot down and saying ‘no AI,’ but offering that they use it in a constructive way,” Luth says. “I feel for our students — they are presented with this technology that is so accessible, so easy to use, and they have so many demands on their time. The temptation to use it is real and not going away. The key, I think, is to emphasize what they are missing out on by not doing the work themselves.”
The Role of AI in Professional Work
That said, Luth addresses the mixed messages that students are encountering about the role that generative AI should play in their lives.
“While the use of AI in education is frowned upon, professional settings are embracing this technology,” he says.
He cites commercials for Adobe and Microsoft, in which rushed employees use the technology to put together last minute meeting agendas, to great success. However, the goal of these companies is not to create efficient working environments, but for a return on their investment in the technology itself.
“There is a huge push to use it, and with it comes an attitude that this is all inevitable, that we have no choice but to embrace it,” Luth says. “I don’t know why we have to give in, if it may not be in our best interest.”
He notes that communication skills are integral, even for someone intent on using AI for their work.
“There is a difference between someone who has strong critical thinking skills and expertise using AI to create a product versus someone who is using AI to avoid the process that will help them learn those things in the first place,” says Luth. “I’m concerned about what will happen to students growing up in an environment where they can off-load so many of these important cognitive processes.”
There is value, Luth says, in doing things the long way. “The speed of AI, the instant gratification it promises, is antithetical to a college education. Struggling, making mistakes, and using careful deliberative thought to figure out a problem — that is how we learn, and that is how we grow.”