Choosing a path on teaching with generative AI

Instructors essentially have three choices as they decide how to approach the use of generative artificial intelligence in their classes:

• Do nothing
• Ease into AI use
• Go all in with adoption of AI

Each of those choices has implications for instructor time, student learning, student motivation, and student trust. It is impossible to predict the extent of those implications with any certainty. We do know that the use of generative AI is growing, is being integrated into daily work life, and is becoming easier to access through common digital tools. Many students are already using generative AI, whether instructors permit it in their classes or not. We also know that AI detectors offer uncertain results at best. They are easy to fool, and they falsely flag a substantial amount of student work. (See We can't detect our way out of the AI challenge.)   

We can’t tell you what to do in your classes. We can help you think through the options and the implications, though.

Do Nothing

Rationale: I’ve already had to deal with enormous changes over the past few years, and the pandemic has left me feeling burned out. Demands on my time continue to increase, especially as many students struggle with motivation, anxiety, and mental health. Keeping classes AI-free is appealing. Neither my department nor the university has offered me time to remake my assignments, and revamping a class takes a lot of time. Besides, colleagues who have more energy and understanding of technology are better positioned to experiment with new AI tools and provide a clearer path forward. The work of those trailblazers will help me become better informed before I make any major changes in my courses. Besides, the world never changes as fast as some people predict it will.

Short-term implications

• Instructors save time by making few, if any, changes in their courses.
• Instructors and many students feel comfortable using familiar approaches.
• Students continue to learn writing, coding, and problem-solving in time-tested ways, unaided by newfangled technology.
• Different classes may have different rules on the use of generative AI, adding a burden to students and leading to potential confusion.
• Many students will use generative AI anyway and look for ways to foil AI detectors.
• Instructors must spend time vetting student work for AI use and, potentially, pursuing academic misconduct cases.
• Instructors rely on the use of AI detectors to flag student work, detectors that have many flaws and that have led to thousands of false accusations.

Long-term implications

• Instructors and some students may grow complacent, falling behind in learning about and using generative AI.
• Instructors may have to spend more time in the future learning about generative AI and changing more aspects of their classes.
• Instructors and students miss out on opportunities to learn about ethical co-production with generative technology.
• Some students will grow resentful about being told to avoid new technology they see as useful and important.
• Instructors miss out on opportunities to rethink and reorient their courses to take advantage of new technology that helps students learn in different ways.

Ease into AI

Rationale: Rarely does everyone change all at once, especially in academia. I recognize the need to change as generative AI spreads and threatens the integrity of assignments and assessments. I know I can’t just ignore that. At the same time, I’m not ready to abandon pedagogical approaches that help students improve critical thinking and other high-level skills. Students may very well be able to plug assignments into chatbots and get adequate responses, but in doing so they will bypass the challenging steps of learning they need to experience. Until I can identify ways to maintain critical skill development in the use of generative AI, I plan to look for a middle ground. I will experiment where I can while maintaining an approach that has served many previous students well for decades. Nearly all change occurs on a continuum, and I’m probably in the middle of that continuum.

Short-term implications

• Students continue to gain skills through traditional ways of writing, researching, coding, and problem-solving but also learn some basic AI skills.
• Instructors gain more time to think about the ramifications of generative AI and how to adapt course material.
• Instructors build out new materials gradually, learning from successes and failures, drawing on the experiences of colleagues, and cutting down on stress brought about by rapid change.
• Use of even some generative AI will require instructors to spend time and attention on discussing ethics.
• Instructors must spend time explaining when use of generative AI is allowed.
• Different classes may have different rules on the use of generative AI, adding a burden to students and leading to potential confusion.
• Many students will use generative AI for all their work anyway and look for ways to foil AI detectors.
• Instructors will spend more time vetting student work for AI use and, potentially, on pursuing academic misconduct cases.
• Instructors will rely on the use of AI detectors to flag student work, detectors that have proved to have many flaws and that have led to thousands of false accusations.

Long-term implications

• Instructors and students miss out on opportunities to learn more quickly about ethical co-production with generative technology.
• Students will learn some new skills, but their understanding of generative AI may not keep up with the needs of potential employers, putting students at a disadvantage.
• Instructors miss out on opportunities to rethink and reorient their courses in more substantial ways.

Go All In

Rationale: AI may very well change education forever. It provides new ways of researching, writing, and coding, and it performs low-level, repetitious tasks with ease and speed. Ultimately, that will save my colleagues and me time and allow us to focus on higher-level skill development. Jumping into AI-aided instruction now will also help me gain valuable experience with tools that are changing many aspects of work and life. Experimenting with AI-augmented teaching will cut down on a constant need to check for student use of generative AI and will allow me to adapt my courses based on evidence of student learning. I know there will be rough spots, but allowing my students to use generative AI will cut down on stress and give me freedom to innovate.

Short-term implications

• Remaking courses and assignments will require time now and will require constant adaptation during the semester as instructors see how students handle the use of generative AI in learning. That may add to instructor stress.
• Instructors must learn and practice with new generative AI tools. That takes time.
• Students must also learn to use generative AI tools, so instructors will either have to set aside class time for that or have students learn to use AI outside class.
• Instructors must address the ethics of AI use through course modules and class discussion.
• Adapting to generative AI now will save time in the long run.
• Instructors learn new ways of engaging students and helping them learn.
• Instructors spend less time tracking student use of AI, easing stress on students and instructors.
• All students will have access to the same tools.
• Some students don’t trust technology and will be resentful of having to learn new AI tools.

Long-term implications

• Students learn cutting-edge skills that will set them apart in the job market.
• Students may miss out on traditional ways of learning and skill development.
• Instructors put themselves at the leading edge of change, helping to shape the use of generative AI in education, and saving themselves time in the long run.
• Instructors use largely untested approaches to assignments and disciplinary learning.
• Generative AI has raised many concerns about privacy and security, and instructors must consider those concerns in class adaptations.
• Generative AI is opaque, and it is impossible to know exactly how it creates writing and coding, or what role it will play in the future.

If you still aren’t sure which approach is best for your classes or if you are ready to move forward, we have many other resources to help you prepare.