When talking to young people about AI, you might expect to see curiosity and excitement, and perhaps some mild confusion. That’s what I expected too, but when I included high school students in my research on emotions in human–chatbot interaction, the strongest emotion I encountered was frustration. “The students here are performing very well with ChatGPT, especially in exams. It’s really frustrating,” a high school student in Germany told me.
The following article explains how our school system puts students at a double disadvantage. Not only are they not prepared for an increasingly AI-driven job market, but they also leave school without the cognitive skills they should have acquired. However, this is a failure of design, not destiny, and I will argue that action is overdue, showing what changes are needed, starting today.
The Double Disadvantage
Following the publication of ChatGPT in November 2022, competing AI chatbot models entered the market, quickly making chatbot interaction a mainstream phenomenon. By 2026, schools still have not adequately addressed the topic, missing an opportunity to prepare young people for life in the AI age.
Missing AI Literacy
In their exhaustive paper on AI literacy, Duri Long and Brian S. Magerko identified that the ability to adapt to innovations and the competence to work creatively with AI is becoming a fundamental professional skill. The International AI Safety Report 2026 states that professional success depends on workers’ ability to use AI to complete their tasks. Despite political recommendations such as those made at the 2024 Conference of Education Ministers in Germany, there is still no comprehensive concept for dealing with AI in schools that has been implemented, nor any AI literacy education that has been anchored in the curriculum. Even existing courses have often not been updated and a student reports, that in 2026 “in fact, none of the computer science courses cover AI”. Consequently, students graduate without understanding how these systems work, what they can and cannot do, or how to use them productively, unless they learn this independently. It is important to note that mere usage of AI does not lead to AI literacy, which requires a deeper understanding of how AI systems function, their limitations, and how to navigate a reality filled with AI-generated content. These competencies are essential for a democratic society based on citizens who can develop informed opinions and participate in debates. Given this, is it fair to place the full responsibility for taking the initiative and investing time in teaching themselves outside of school hours on young people?
Students report that some teachers are currently using AI without a clear framework in individual exercises. “There are some teachers who just give assignments and say, ‘Do this and that with ChatGPT,’ because they want us to practice how to use ChatGPT and stuff, but mostly they don’t.” However well-intentioned, the lack of expertise and competence among teachers often has the opposite effect, as another student describes:
“And so they try it, but it just doesn’t work because the students are 17, 18, and we don’t go along with it. I don’t get involved in it either, because I don’t see the point in teaching it to the teachers, because I know how it is. I’m not completely stupid, after all. Um, so they try it. It’s sometimes actively incorporated, but it’s actually used very deliberately by us students. […] We do our thing and that’s just how it is.”
This is an institutional failure, not a personal one. Teachers are having to navigate this rapidly changing technology without support. There is no actual curriculum or larger plan in place to educate people on the different dimensions of the topic.
Uncontrolled AI Usage
Fact is, AI is used in schools. But it is often used unproductively and without control, by both students and teachers alike. Students use it to bypass annoying schoolwork. And who could blame them? According to the interviewees, the vast majority of students use AI to complete homework, write essays and prepare for oral presentations, despite it being officially forbidden.
“Yes, I think […] everyone uses it. The teachers know about it too. They always say, ‘We’ll notice if you use it.’ Yeah, they don’t notice. They really only notice it very rarely, or they just don’t want to deal with it. […] [Students] get by very well with ChatGPT, especially in exams. […] there are people sitting there with their smartphones out, just taking pictures and copying from ChatGPT.”
The use of AI is not permitted, but it is often impossible for teachers to detect it reliably, as it is genuinely impossible to tell the difference. This leaves teachers with no option but to either ignore it or guess blindly. Neither of these seems like a good or fair solution. Ignoring it suggests that the school is unable to control the situation. Guessing blindly can result in the wrong people being punished, leaving students who do not use AI feeling frustrated.
Additionally, if not made transparent and detected by the students, the use of AI by teachers to create class content sets a undermining and negative example. A student describes how she and her maths tutor discovered an AI-generated question in an exam. “At first, I thought, ‘That can’t be true, he would never do that, he’s a teacher.’” But the question was not only AI-generated, it was also unsolvable due to logical flaws. Such negative experiences can teach the wrong values and risk fostering a negative attitude towards AI and school, preventing students from developing an interest in learning about and engaging with AI in their free time in order to build relevant skills.
This situation does nothing to help students develop the skills needed to collaborate with AI systems, and it undermines the very purpose of schools. Perhaps the most troubling consequence is that students are missing out on opportunities to develop core cognitive skills as all this bypassed schoolwork is exactly the kind of work that, until recently, helped to develop them. If these tools are no longer effective, they need to be replaced with new learning concepts.
Why It Matters Beyond The Classroom
The consequences of this dynamic extend far beyond school grades. Students who have not developed strong cognitive skills will struggle to critically evaluate AI outputs, making them more susceptible to misinformation and less capable of using AI effectively. Students who did not get to know AI as a productive collaborator at school and who carry frustration from that experience are unlikely to invest personal time in building important AI skills. The result is a generation entering adulthood in the age of AI that is unprepared on both fronts. This is not an individual problem. The effects reach into the labour market, public discourse and the long-term cognitive health of a generation, thus affecting the core of society.
The labour market is changing rapidly. This poses a challenge to society and carries risks for democracy, depending on the speed and extent of the changes, as well as how they are managed politically. Young people will be disproportionately affected, as AI will initially replace entry-level jobs. Studies from Denmark and the USA have already found a decline in employment for early-career workers in AI-exposed jobs since 2022. In the UK, firms with a greater exposure to AI have slowed down hiring, particularly for entry-level positions. Preparing the workforce for an uncertain AI-driven future requires more than just technical skills; it requires adaptability — the capacity to learn, unlearn and relearn as technology evolves. Students who leave school without AI literacy or curiosity about AI will enter the job market at a significant disadvantage. They lack the necessary technical expertise and the motivation to acquire it. This can lead to high unemployment rates, as well as anxiety and frustration amongst young people.
The cognitive dimension is equally concerning. Cognitive offloading — the delegation of mental effort to AI — can result in a weakening of critical thinking and memory, as well as in a discouragement of active reasoning and verification, and in a loss of skills in general. This is not merely an abstract concern about intellectual potential. Weakened cognitive skills have practical consequences. Those who have not learned to think critically are more susceptible to misinformation and disinformation — precisely the kind of AI-generated content flooding digital spaces. They are less equipped to evaluate AI outputs, recognise errors or bias, and maintain the healthy scepticism that responsible AI use demands. In a society where AI-generated content is becoming ubiquitous, cognitive resilience is a necessity. The long-term consequence of a degeneration of these cognitive skills could be an erosion of individual autonomy. While this affects everyone using AI, teenagers, whose brains are still developing, are particularly vulnerable.
The current state of education contains ingredients that have previously been shown to undermine social stability and democratic societies: anxiety, frustration, susceptibility to misinformation and the erosion of autonomy. Dealing with these ingredients before they fully unfold, neutralising them and perhaps even transforming them into something beneficial concerns society as a whole and must no longer be neglected or postponed.
What Needs To Change
The above problems are serious, but not inevitable. They are the result of institutional inaction and can be resolved through deliberate, structured intervention. Three priorities stand out, each operating on a different timeline.
Immediately: Train the Teachers
Nothing changes in the classroom without the teacher. Currently, most educators are navigating AI with little guidance or training, even though they need to be equipped with three core competencies.
1. The ability to identify tasks prone to AI offloading. 2. The ability to incorporate AI productively and actively into their teaching. 3. The creativity to design tasks and assessments that genuinely engage students’ cognitive abilities and cannot simply be outsourced to a chatbot.
This also means rethinking the concept of homework in the AI age. Traditional take-home assignments, in which a student works alone with a prompt and a deadline, are the format that is most vulnerable to AI substitution. Oral examinations, live problem-solving exercises, Socratic classroom discussions and collaborative projects carried out in real time provide a better assessment of the skills that matter most. Teacher training programmes must support educators in making this shift in both principle and daily practice.
In the Short Term: Create Positive AI Experiences
Schools need to create spaces where students encounter AI as something genuinely exciting, rather than as a source of unfairness or confusion.
Project days and hackathons, which only take a few days per school year, can be powerful interventions. They provide dedicated time for students to build something with AI and experiment. These are not a replacement for structured AI literacy education. Rather, they complement it and serve a specific purpose: reigniting curiosity. Students who spend a day building a chatbot, generating art or prototyping a solution to a real-world problem using AI tools are far more likely to continue exploring AI in their own time.
Concepts for these project days can be developed in cooperation with NGOs specialising in technology education. There are organisations across Germany and Europe with deep expertise in making technology accessible and engaging for young people. Collaborating with them — and drawing on international models — would accelerate the development of these initiatives in terms of both quality and reach.
Systemically: Make AI Literacy Part of the Curriculum
The long-term goal must be structural. Understanding what AI is, how it processes data, its limitations and how to use it responsibly needs to be a regular, mandatory part of the school curriculum from elementary school onwards.
Research from Finland highlights why it is important to start early. A study by Mertala and Fagerlund involving Year 5 and Year 6 students found that those lacking basic AI literacy were significantly more likely to anthropomorphise AI, believing that it can think, feel or understand them. This is not a harmless misconception. It increases trust in AI outputs, fosters emotional dependency and can lead to parasocial intimacy with chatbots. Teaching young people how data processing actually works is a protective measure against these effects.
In Germany, this responsibility lies with the state-level ministries of education. These ministries must develop structured AI literacy programmes and integrate them into the weekly schedule as a core component of education, not as optional enrichment. Finland’s national AI literacy strategy offers a valuable model, and peer exchange within the EU could significantly accelerate development. This is not a challenge that each country, or even each federal state, needs to solve independently.
A Question of Societal Resilience
The way in which we educate young people about AI today will determine the kind of society in which we will live tomorrow. If we fail to equip young people with the skills to use AI and the cognitive tools to think critically about it, they will struggle to participate meaningfully in an AI-shaped economy and democracy.
This is not just an educational issue. It is a question of societal resilience. In order to build a democratic society that can navigate the AI revolution, we need citizens who understand the technology, can think independently and approach it with informed confidence, rather than frustration or blind trust. This process must begin in school.
The tools to effect this change already exist. We are beginning to understand what works. What is missing is urgency. The students I interviewed are about to leave school and are already experiencing the consequences of inaction. We cannot afford to wait.
Note: all student quotes were originally in German and have been translated while maintaining the original tone and content.
