How do we teach AI literacy when students already think they’re experts?
Created in partnership with
Created in partnership with
You may also like
I recently asked my students to redesign a Moroccan road safety campaign using generative AI.
They began work quickly and within an hour they had produced something that looked genuinely professional, with bright visuals, sharp messaging and even hashtags that felt ready to go straight out on to social media.
Then I asked them a question: “Where did these images come from, and do you actually have the right to use them?”
No one answered immediately. A few looked back at their screens, and one or two tried to guess.
But mostly there was silence. This moment told me more than the campaign itself.
When confidence charges ahead of understanding
Students today aren’t intimidated by AI. If anything, they’re ahead of us in terms of how quickly they experiment with it, try things, remix them and move on.
But this confidence can sometimes be misleading. Yes, they know how to get results but they don’t always understand how these outcomes are achieved or whether there are any associated problems that will need further exploration.
- AI literacy is the bridge between fear and the graduates we need
- Reaching for the pizza? How to help students make informed choices with GenAI
- Why your AI training programme won’t create an AI-literate university
When we discuss AI literacy, this is a gap I keep returning to. Not if students can use the tools effectively but whether they can pause long enough to question them. And this point is only going to become more pressing.
The students we’re teaching now, Gen Z, already treat AI as everyday normality. The next cohort, Gen Alpha, will arrive having grown up with it in ways we are only beginning to understand.
So the question won’t be whether they can keep up with the technology. It’s whether we are helping them fully make sense of it.
What changed when we slowed down
This latest semester, I tried something slightly different. I introduced a course called Digital Media in the Age of AI. On paper, it looks like a typical media course but in practice it became something else: a space where students had to explain their decisions, rather than simply produce outcomes.
We still used AI tools throughout but what changed was what happened after the output appeared.
Instead of moving on to the next thing, I asked them to stay with it and consider where the results might be biased, what assumptions were built into the AI and if this would make sense to a Moroccan audience or if it felt imported.
At first, this approach slowed them down in a way they didn’t like. They were used to moving quickly and producing something, submitting it and receiving feedback later.
This was quite different. The process asked them to question their own work while they were doing it.
Ultimately one idea that seemed to stick is something I’ve called “AI sovereignty”. This isn’t a technical term, I just used it as a way of asking: where is your data going, and who owns it once you’ve entered it?
For many of them, this was new territory.
They had never really thought about what happens behind the interface, or considered who was storing their inputs, how this data might be reused or what they were giving away in exchange for convenience.
Once that clicked, it was possible to see a real shift in mindset. They didn’t stop using AI but they started treating it more carefully.
Some of the simplest activities ended up being among the most effective. I began asking students to run the same prompt through two different AI tools and then compare the results.
When they tried this with Moroccan cultural content, the differences were obvious and significant. Certain outputs leaned heavily on clichés. Others missed key details entirely, such as important language, symbols and context.
It quickly became clear that AI is not neutral. Rather it reflects what it has been trained on and what it has not. This realisation perhaps achieved more than any lecture I could have given.
Less polished but more engaged
Early in the semester, I gave students complete freedom in some of the labs. The results looked impressive, polished, creative, sometimes even surprising.
But when I asked them to explain what they had done, many answers were thin. So we adjusted further.
I started asking for short explanations alongside each project. Nothing long, just a few sentences about how they had used the tool, what they trusted, what they didn’t and why.
Long reflections didn’t work particularly well. Students wrote what they thought I wanted to hear. Short prompts worked better, in that they were more direct and honest.
Some of the most interesting moments happened when the class moved beyond the assignments. Students organised their own discussions and short webinars on topics such as AI in elections, misinformation and creative work.
These sessions felt different from typical presentations in that they were less polished but more engaged. Students felt comfortable challenging each other. They brought in examples and disagreed openly.
That’s when it became clear that AI literacy is about more than technical skill. It involves conversations that talk through the implications of what these tools can do.
Local context mattered more than I expected. At the beginning, many student projects felt disconnected from Morocco. The visuals used could have come from anywhere and the tone was often generic.
It took time, and feedback, for students to start adjusting their work. They began paying more careful attention to language, imagery and audiences. These were small details, but important and valuable ones. By the end, the difference had become noticeable. Work felt more grounded and intentional.
What this might look like elsewhere
I wouldn’t describe any of this as a finished model. It’s all still evolving. But a few things have stayed with me, and might be useful to others trying something similar:
- Ask students to compare outputs, not just generate them. That’s often where the real learning starts.
- Avoid letting one tool dominate. Using more than one changes how students think about reliability.
- Keep reflection light but regular. A couple of honest sentences can be more revealing than a long essay.
- Introduce questions about data and ownership early. Students rarely think about this unless prompted.
None of this requires a major course redesign. In my case, it was more about shifting the rhythm to build in small pauses where students had an opportunity to think a little more about what they were doing, not just how quickly they could do it.
I don’t think our role is to slow students down. But they can benefit from help to develop a habit of questioning things more, especially when the tools being used seem to work so well on the surface.
If we can do this, AI becomes something they will engage with more thoughtfully. And that, I think, is where real literacy begins.
Abderrahim Agnaou is associate professor of communications & rhetoric and chair of the General Education Committee at Al Akhawayn University.
If you would like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.
