How learning works in the age of AI
Dr. Iemmolo, how does learning change when AI tools take over parts of or even the entire creative process?
I don’t regard AI as an enemy but as an innovative tool. That’s why the tool itself is not the problem but the way in which we use it. Because when, for instance, we’re writing we’re thinking and communicating at the same time. Writing is not the final product of thinking – it’s the act of thinking per se. As soon as AI tools take care of working creatively for us that may be efficient but the cognitive friction that’s necessary for true comprehension is missing. Learning is like a workout at the gym: you need to sweat a little.
The expert
Dr. Giorgio Iemmolo is a linguist and director of the Language Center of ETH Zurich and the University of Zurich. In “tomorrow,” he explains why AI is neither an enemy nor a cure-all – and why writing by hand may be our last bulwark against thoughtless learning.
Where do you see meaningful efficiency gains through AI – and where does the loss of skills begin?
AI can help initiate ideas, structure texts, or serve as a mirror. But when we use it as a substitute for thinking we’re going to lose the ability to reflect on things. AI texts are often coherent but flat – they’re based on statistical probabilities, not on cognition. When users simply adopt these texts, they’re going to lose the ability to actively link language and thinking.
What methods do you recommend for integrating AI into the learning process in meaningful ways?
I have my students write with AI, for instance – but only to analyze what’s lacking. They should reflect on why a text sounds good, but remains hollow in terms of content. That meta skill of recognizing the difference between well-sounding and sound reasoning is vital.
What exam formats promote independent thinking despite AI?
I’m working on a more oral and process-oriented design of exams. In Italy, where I went to university, students orally defend almost every exam, which forces them to reflect. I’d like for students to not just turn in their work but be able to explain it: Why did I write this like that? What sources did I use? That kind of self-reflection is a lot more conclusive than a finished PDF file.
“AI is a tool but thinking, comprehension, and responsibility – they’re going to remain human."
Dr. Giorgio Iemmolo
What role does writing by hand still play today?
A truly major one. Many mathematicians or engineers at ETH Zurich still write their formulas by hand although they’d long be able to use digital tools. Why? Because when writing you decide what’s important. Something that’s handwritten forces you to think about it – and that’s exactly what we need in terms of linguistic expression as well.
Should school and college students really continue writing classic homework assignments?
Yes, absolutely – but with one difference: We should more intensively assess the process, not just the product. A good homework assignment is created in stages: research, reflection, writing, revising. That includes handling AI too. School and college students could document their prompts, reflect on their utilization, and show what they’ve learned from that. It’s important that teachers learn to recognize and assess those processes.
How great is the risk that we lose the ability to understand what we write when we rely too much on AI?
I do see that risk. Anyone who strictly generates without personally reflecting is going to lose the ability to consolidate knowledge. Neuroscience supports the notion that writing equals thinking too: during the process of writing neural connections are formed that are crucial for abstraction and long-term memory. Especially people who write by hand activate regions in the brain enabling deep learning and thinking in terms of concepts. Writing helps the brain recognize major correlations and to form specialist knowledge that not only feeds on facts but is based on comprehension and can be used in many different contexts. Research pertaining to automation knows that cognitive systems atrophy when intellectual tasks are outsourced to machines. Plus, there’s a second stumbling block: the products of generative AI skillfully imitate knowledge, thus covering up our dwindling skills. We sound eloquent without truly understanding and without realizing that we don’t understand.
Should we introduce “AI-free learning phases”?
Yes. Absolutely. AI-free phases create an awareness of one’s own thinking. That may be a handwritten exam at school or university, an oral defense, or short reflective texts in which school or college students capture their learning process. Important in all areas of life is for us to build knowledge, for room to exist for imperfection – for the friction from which cognizance emerges.
How do recipients, whether teachers or HR departments, for example, detect if a text has personally been created by the sender?
Honestly, AI detectors hardly work. I prefer relying on conversations. Five minutes of discussion tell me more about a person’s understanding than any software. If someone can’t explain their own text, I know enough.
"We need to learn to re-appreciate the old virtues of the humanities. Reflection, reasoning, tolerance of ambiguities."
Dr. Giorgio Iemmolo
What skills do we need in the age of AI?
Judgement above all. For me, AI skills don’t mean being able to handle any tool but to know when and what I use it for. Linguistic precision, critical thinking, communication skills – those are the abilities that are becoming increasingly important. Plus, we need to learn to re-appreciate the old virtues of the humanities. Reflection, reasoning, tolerance of ambiguities – they’re the foundation for any technical future.
Are you afraid of the next AI generation?
No, I’m not afraid of AI. I’m sometimes afraid of the discourse about AI – when it’s portrayed as a savior or as a threat. Both are wrong. We should neither become euphoric nor hit the panic button. AI is a tool but thinking, comprehension, and responsibility – they’re going to remain human.
