What do you think of AI?

Are you an AI Enthusiast/Optimist? Or are you more of an AI Realist? Or maybe you’re a member of the AI Resistance? Whether we love or hate it or fall somewhere in the middle, AI in general and generative AI in particular are here to stay. When thinking about getting started with AI and using AI in the classroom, how can we discourage or prevent its use in ways that would be considered academically dishonest? Finally, how can we ensure that these tools do not harm or inhibit our students’ learning and intellectual growth?

Does “AI detection” work?

One way some faculty are attempting to combat the use of generative AI is through AI-checkers. These programs claim to be able to detect whether a sample of writing is AI or a student’s own words. The accuracy is debatable. They can be outsmarted by swapping words for synonyms and “massaging” AI-generated text, so students could do–and are doing–the same. Also, students with more “predictable” writing–that is, writing that has low perplexity and more steady sentence structure and length–are likely to have their work flagged as AI-generated. This may disproportionally affect English language learners and neurodiverse students. Even one student wrongly accused of plagiarism-via-ChatGPT is one too many. There are also student data privacy and intellectual property concerns with feeding student work into AI detection or gen-AI algorithms or programs.

What can educators do about it?

First of all, why might a student use generative AI, such as ChatGPT, to complete their writing work for a class? For some, they might be committing academic dishonesty and for the same reasons they might commit other types of academic dishonesty: assignments that students see as busy work, demotivating, uninteresting, pointless, or formulaic; fear of failing; procrastination; stress or time management difficulties; and lack of understanding either about what constitutes academic dishonesty or the consequences thereof. On the other hand, students may be using generative AI to improve their learning and streamline their studying and writing processes by having AI proofread their work and give them suggestions for improvement (AI-as-tutor), generate ideas for a project or paper, or generate study guides or practice exams for tests and quizzes.

As educators, we can address some of the academic integrity issues ourselves. First, we can clearly explain (and reiterate often) what we consider academic dishonesty in our classes and what consequences will be and then remain consistent in enforcing consequences. We can get to know our students as people and the demands on their time so we can help guide them to resources that may help with stress or time management issues. We can scaffold assignments so that procrastination becomes less of an issue with more structure built into the process. We can give students opportunities to demonstrate their learning and growth throughout the semester rather than with only high-stakes tests in the middle and at the end of the course that are worth a significant portion of their grade. We can also encourage students’ ethical use of AI to enhance their learning by encouraging certain types of AI interaction.

One other important thing we can do is rethink our course assignments. In other words, make them AI-resistant. We won’t attempt to make them “AI-proof” because that may not be possible, at least, not for much longer as generative AI gets better at mimicking human output. What we can do is think critically about the kinds of assignments we give to students and consider why students might use generative AI to complete them instead of or in addition to doing the work themselves. Maybe they’re using ChatGPT because our assignment is boring or students don’t see the value in it or it isn’t meaningful or students see it as busy work… Or students might be using ChatGPT for similar purposes that they used other resources in the past; for example, searching the internet for topic ideas, outlines, study guides, or practice tests.

What are some specific examples? 

How can we assess students’ learning and progression towards learning outcomes in more meaningful and engaging ways? What assignments can we create that students will actually want to do? These are the questions each of us can ask ourselves as we take on the exciting but admittedly daunting task of changing our approaches to teaching and assessment in the age of generative AI.

AI Policy

Have a clear-cut AI policy for your course. You can find a few examples here. Even better, collaborate with students on an AI policy for your class so that they can learn more about and provide input into the ways AI should and shouldn’t be used in your particular course and discipline. Giving students some sense of ownership over creating the policy may create more buy-in.

Experiential or Project-based Learning

Use experiential or project-based learning to give students the opportunity to learn by doing. Students will be able to connect what they learn in the classroom more meaningfully to the real world. Hands-on activities, field trips, service learning, community-based research, and field work are all examples of experiential learning. You could also assign more complex, multilayered projects with pieces that build upon one another (scaffolding). For example, students could design an infographic, record a mini-podcast episode, produce an interactive timeline, or create a public service announcement.

Active Reading

One of the age-old quandaries is how to get students to read for homework. One way is to teach students to annotate their texts (print and digital) while they are reading, and have them submit those annotations for credit. You could also, consider collaborative annotation where multiple students annotate on the same document. Some tools for digital annotation include Perusall and Hypothes.is, but you can also have students annotated printed texts.

Discussion Board Alternatives

Traditional discussion boards can be rife with AI-generated content, particularly if students don’t see the value in spending the time and effort on them or see them as busy work without any real application to a major assignment, for example. You may consider using live chat tools for brief scheduled class discussions instead of traditional discussion boards to create more immediacy in a conversation. You could also use tools like whiteboards (physical or digital) to promote non-linear discussion/dialogue in your online or in-person course. (Just make sure to check on the accessibility of any learning tool you decide to use.)

Metacognition

Metacognition, or being aware of and understanding your own thought process and can strengthen student learning and academic performance. There are numerous ways to implement metacognition i your course, depending on the course and discipline and your objectives for student learning. One example is to have students explain their thinking or process on a project, performance, or writing assignment (i.e., a sort of self-annotation). Another option would be to have students synthesize insights they gained from live class discussions and use those as study guides for an upcoming exam, notes to turn into an essay, or topic ideas for a class presentation.

Keep It Local

Generative AI may be more limited in its “knowledge-base” when it comes to hyper-local issues. To take advantage of this, you could have students research or engage in hyper-local issues, within our own school community, for example. This has the added benefit of getting students involved locally and showing students how what they are learning in the classroom is relevant to the real world.

Citing Sources

Finally, as of right now, generative AI isn’t great at in-text citations and references. It can generate them, but they may be incorrect or “hallucinated” (a.k.a. made up). If you require in-text citations and references for your students’ work, make sure to verify them for accuracy, existence, and relevance.

Final thoughts

What makes generative AI alarming to us as educators are the ethical concerns, yes, but also the impact its efficacy and accessibility has had and will have on our profession and students’ learning. Will AI replace us all? No, but it will inevitably reshape the way we teach and students learn.

Two things we have that AI doesn’t is emotional intelligence and human connection, so let’s use those strengths to enhance our students’ learning, assess and meet our students’ needs in real time, promote critical and creative thinking, and increase student motivation and sense of belonging. Let’s use AI as the impetus to revitalize and inspire our teaching and assessment approaches.