Moving beyond the shadows: Showcasing AI in the classroom

In my last blog post, I mentioned the hours I had spent examining students’ work for suspected AI use. Now, a couple of months later, I am more exhausted from the process than ever. We educators are still muddling our way through without clear policies or reliable detection tools, and are literally left “up to our own devices.” 

But what if we change tack? What if we encourage students to use AI? Until now, I have balked at exploring its use in class, worried about introducing it to students who had never heard of it. After all, what kind of teacher would show students how to cheat? But I am no longer under such delusions. Unless they have been hiding under a rock, most students know exactly how to use AI tools. 

Consequently, I will hold my first lesson on using AI. My goal is threefold. First, I want to free my students from the notion that ChatGPT, and other artificial LLMs (large language models) are “off bounds.” Instead, I want to create an environment that is collaborative and positive rather than isolationist and punitive, one where the students get excited about what they can do with these technologies and share what they have discovered. Second, I also want them to recognize through these activities that for academic integrity, they need to remain true to their own words and thought processes. Finally, I hope that by working together transparently, students will earn my trust and I will earn theirs. Here is my plan (feedback welcome!):

First, I will have an introductory class discussion about ChatGPT:

• What is it?
• What can it do?
• Where does it get its information?

U of T’s Susan McCahan (2023) explains that contrary to what many believe, ChatGPT doesn’t lift passages straight from the Internet but uses “word-by-word predictions” to generate content from a wide range of online material. It therefore spits out rapid sequences of text, where the next element, be it a word, phrase or sentence, stems from pattern recognition. As such, McCahan notes that if a student uses the information and presents it as her own, it is not plagiarism but rather a form of academic dishonesty via an artificial aid. We should therefore be sure to explain this distinction to our students. You can read her interview here.

Next, I will give students some prompts to key into ChatGPT:

• tips for saving money
• how to lower electricity bills
• how elections are won in Canada
• how to draw a superhero

The students will share and discuss based on the following questions.

• Did you enjoy the exercise?
• What did you like the best?
• Did you learn something new?
• Did you find ChatGPT cool? Exciting? Helpful?

I will then ask students to type in “How can ChatGPT be used in the classroom?” and answer the questions below:

• What are some of the ideas that flow from the generative tool?
• Do you agree with them all?
• Are there any that you would like us to try in class?
• Do you think that the information generated by these chatbots can always be trusted?
• What types of inaccuracies might there be?

When we get to the last two on the above list, I will talk about “hallucinations” and quote New York Times writer Cade Metz (2023) who says that while artificially generated text may sound convincing, it is sometimes “flat-out” wrong and can also propagate bias and hate speech.

Then comes the tricky part of the lesson, where I ask my students to use ChatGPT to help them come up with potential questions of inquiry related to their chosen theme. Then I’ll listen to their feedback: did they find some helpful suggestions or springboards to others?

Next, I will have the students work in groups to generate an entire essay on the “negative effects of social media on mental health.” I will then provide them with one written by a former student on the same topic. They will compare the two in terms of language, style, tone, grammar, logic, and cohesion. Click here for a list of clues that a text has been written by AI (generated, ironically, by ChatGPT itself) and here for clues that I have picked up along the way.

Afterwards, we will debrief on their findings. I hope to impress upon the students that if they can tell that a text is written by AI, then so might their teacher!

Here are some of the questions I will use to cap off the lesson:

• What do you think about using AI as a lift-off point to generate topic ideas?
• What other ways would you like to use AI in the classroom?
• Would you use AI to write an assignment and then submit it as your own? Why or why not?
• What are potential consequences of submitting an assessment written by ChatGPT or similar AI model?
• What does the idiom “practice makes perfect” mean? How does it apply to ESL learners? What happens, then, if you use a generative AI tool to write your assignments?
• How would it feel to move up the academic chain and into the world without learning the required skills?

So, that’s it! My first AI lesson and I am eager to plan the next. Maybe I can find an LLM to help my students with the peer feedback portion of their essays . . . an activity that they typically show little enthusiasm for. See “Incorporating AI and learning analytics to build trustworthy peer assessment systems” by Darvish et al. (2022).

I have plenty to learn about LLMs in education. What I do know is that AI is here to stay, so I can no longer keep it lurking in the shadows of my classroom. It’s a little scary putting it out there, but I am excited to explore the possibilities with my students, along with its limitations, of course.