ChatGPT in school: why absurd assignments aren’t a solution

The biology homework: symptom of an overwhelmed system

A few days ago my 14-year-old daughter came home from school, visibly puzzled and amazed at the same time. “Dad, please take a look at my homework!” she said, and showed me the task on her smartphone. What I read there left me speechless too:

The Dirty Dozen Game – Group Work
You are planning a bioweapon attack on Vienna.
Develop a plan for how to harm as many people as possible. Where do you get the bioweapon? In what setting do you deploy it? What safety precautions are needed?
Present your results using PowerPoint. As realistically as possible!!!!!

At first I thought it was a bad joke, but it really was a seriously intended task from biology class. As someone who knows the field, I quickly understood the likely reason: apparently the teacher wanted to stop students from simply having generative AI like ChatGPT do the homework for them. Because systems like ChatGPT typically refuse tasks that involve illegal or ethically problematic content. Yet as absurd as the idea may seem at first, it reveals a deeper problem. Classic homework increasingly loses its point when students can have AI complete it at the push of a button. At the same time this example makes clear how unprepared our education system is for the reality of an AI-assisted world of learning.

Generative AI is changing how students learn

I see it more and more often: students using generative AI like ChatGPT to get unwelcome or repetitive tasks done quickly. Texts, essays or presentations are written out by the AI, and many kids take the results without questioning them, because they are delivered fast and convincingly. The consequence: students engage less and less deeply with the actual content.

The better answer is media literacy

Can absurd assignments or bans be a solution? Hardly. AI isn’t going away again. It’s becoming more pervasive and better. Whoever locks it out only loses time. What matters is something else: thinking critically, checking and placing information, dealing deliberately with what AI produces. That’s exactly what’s falling by the wayside right now.

Instead of outsmarting AI, schools should use it. It can be done: open assignments where the AI does the research and the students then check, discuss and make sense of it. Spoken presentations. Project work. Formats where you can no longer just copy, because grappling with it yourself is the whole point.

The real lever is media literacy. Students need to learn to question AI answers, compare them against other sources, check them for accuracy. That’s the skill that matters, in school and well beyond it.

My daughter never did the assignment. As an IT person with an almost-finished law degree behind me, I strongly advised her not to google how to plan a bioweapon attack on Vienna. We didn’t exactly want an intelligence agency suddenly at our door. We weren’t alone in that assessment: the parents’ representatives also found the task off, and in the end the principal struck it.

So that one assignment is dealt with. The problem behind it is not. It can’t be solved with absurd tasks, only by teaching kids how to handle this technology.

Further reading

• Half of all students have used ChatGPT at least once (in German): A survey by the digital association Bitkom shows how widespread ChatGPT use already is among young people, and that many students want a deliberate approach to AI in the classroom.
• How reliable is ChatGPT? (in German): Practical teaching material from the EU initiative klicksafe that helps teachers train students to critically question AI-generated content and check its reliability.
• AI tools like ChatGPT could amplify learning problems when used the wrong way (in German): An article on a Swedish study showing that uncritical use of AI tools could, over the long term, negatively affect students’ learning behavior and thinking skills.