Navigating the Challenges of Teaching AI Responsibly in Classrooms

A panicked email from a colleague, “Can we really not detect AI use? What counts as plagiarism? What should I have said in my syllabus?” captures the uncertainty many professors face. As artificial intelligence (AI) becomes more sophisticated, it is disrupting traditional writing assignments. Two years into the generative AI era, data show that students are widely using AI, but professors are still cobbling together policies with limited guidance from their institutions.

In a survey at the University of California, Davis, 70 percent of undergraduates reported wanting more instruction in AI use, yet only 33 percent of instructors had incorporated AI into their courses. Similar patterns emerge in my own surveys, where 40 to 60 percent of participants use AI tools weekly for writing tasks, but many feel unsure about their effectiveness.

A major concern is that AI use can interfere with learning. Research shows that pilots who rely on autopilot struggle during emergencies, and GPS users lose navigation skills. A large-scale study comparing high school math students using unrestricted ChatGPT, AI-based tutoring, or only textbooks found that while AI groups did better on practice problems, subsequent testing without AI showed different results. The AI-tutored group’s advantage disappeared, and the unrestricted ChatGPT group performed 17 percent worse than controls.

Despite these risks, an outright ban is impractical. AI use is difficult to detect reliably, and banning it drives usage underground. Moreover, students need training to use these tools effectively in their future careers. Professors have a responsibility to experiment with AI and guide students on its appropriate use.

In my graduate-level scientific writing course, I allow AI use but with strategic restrictions. The course AI policy, stated in the syllabus, explains the approach to AI use, transparency requirements, and data privacy protocols. Assignment-specific guidance is also crucial. For example, I allow AI to check grammar, analyze the structure of other aims pages, and evaluate adherence to a checklist. However, I prohibit using AI to generate research ideas, suggest alternative hypotheses, or construct experimental designs.

To help students understand the importance of independent learning, I use a simple analogy: If you want to lift a heavy barbell, you can either bench press it or use a forklift. But to build physical strength, you must lift the barbell yourself. Similarly, your brain needs to engage in the task to develop the necessary skills.

Reflecting on my colleague’s email, I realize their questions reflect deeper anxieties about the changing educational landscape. Rather than focusing on detection and punishment, it’s essential to provide guidance on effective AI use. This includes communicating which thinking skills remain essential and designing assignments that integrate AI meaningfully.

However, the challenge is not uniform. Different educational environments will require strategies tailored to their students’ needs. At the graduate level, motivated students are open to guidance, but at other institutions, generative AI technologies can cause significant issues, leading to students graduating “essentially illiterate.”

In the long term, determining how best to use AI as a learning tool may require a fundamental rethinking of pedagogical approaches. Educators and administrators will need to grapple with the purpose of higher education in an AI-saturated world and how to create a culture that values genuine learning over transactional credentialism.

AI poses immense challenges but also offers a chance to reassess what we teach, how we teach it, and why it matters. With an open mind, a commitment to students, and a willingness to adapt, we can emerge with an education system that is more resilient, relevant, and better prepared for the future.