The goal of this assignment was to identify a problem with AI in our current society and create an alternative vision for how it could be.
The rise of AI is changing the classroom drastically, although it does not seem entirely clear exactly how. The popular press is overflooded with articles about generative AI in education, some encouraging us to embrace it, others warning about the dangers, but everyone seems to agree that our current education system needs to catch up to the current developments (Akyol, 2026; Bardoel, 2026; De Kort, 2026; Duin, 2026; Laarhoven, 2026; Last & Saçan, 2026; Thies, 2025; Timmers, 2025; Van der Valk, 2024; Van der Vleuten, 2026; Van Spronsen & Halfhide, 2025; Westhreenen, 2025). The two main topics seem to be how teachers can incorporate GenAI and how students are using GenAI to cheat.
It is clear a sociotechnical drama is playing on the stage of education. Students are using GenAI chatbots such as ChatGPT to help them with their homework or even do it for them. There are students who use GenAI for finding sources or creating practice exams, while others are using them to outsource their thinking and write entire essays for them. Some are even using it during exams (Bardoel, 2026). Teachers are scared they are losing control (Timmers, 2025). They are trying to reclaim their power by adapting their homework and classes; take-home written assignments are hardly graded anymore, all graded work is written in class under supervision and big projects are starting to become more focused on the process than the end result (anonymous participant, personal communication, May 7, 2026). A number of people are advocating for a change in examination, with less written exams and more oral exams (Bardoel, 2026; Van Spronsen & Halfhide, 2025).
This moral panic, where the new technological developments seem to threaten the traditional education system, is not new. Many news sources compare this current AI drama to the invention and classical acceptance of calculators (Ghai, 2026; Westhreenen, 2025; Vleuten, 2026). These people counter the view of the technological defeatist with the prediction that GenAI will be incorporated as a tool into classroom just as calculators were and that they will prove to be more beneficial to education than detrimental.
In short, there is a lot of uncertainty on what the future will look like in regards to GenAI in education and many teachers are looking at their institutions for guidance in the form of policies, but current research does not have a definitive answer for which policies would be most effective for implementing GenAI into high schools – if it should be implemented at all.
#
A current line of critique regarding current AI development is the ubiquity, availability and accessibility of unguided and unregulated GenAI. In an educational context this critique translates to the fact that generative AI systems are becoming deeply embedded in education before schools, teachers, and institutions have developed effective policies, norms and regulations for their use. GenAI is ubiquitous through its integration into daily online activities. One interviewed student noted you can hardly escape GenAI, as even googling something will provide you first with a GenAI result (Google, n.d.). GenAI is also readily available, as it only needs internet access and will provide answers instantly. It is also easily accessible, as the interface is very intuitive, you simply type in a question or demand and the GenAI will respond, regardless of grammatical errors or difficulty of the prompt.
Unguided GenAI is, for example, chatbots like ChatGPT. Guided AI in this context refers to GenAI systems that incorporate pedagogical safeguards designed to guide student reasoning through structured inquiry (Vendrell & Johnston, 2026). Rather than accepting the instant answer, students are encouraged by the AI to question, compare and revise ideas. Studies have shown that guided and regulated AI use can be very beneficial for students, while unguided and unregulated AI use is often harmful (Bastani et al., 2024 & Basiouny, 2024). Students may become dependent on AI rather than learning the underlying material and perform worse when the AI is removed.
Due to the constant accessibility and ease of access, there are hardly any barriers for students to use GenAI daily. This can be highly problematic, as it means students can be using GenAI without a second thought. Both the students I interviewed admitted to becoming “lazier” by using chatbots. It is now easier than ever to cheat on homework, and there is no guarantee for students who use GenAI wisely, that the information given is correct. GenAI has been known to hallucinate, which means factually made up or wrong text is provided authoritatively (Boretti, 2026). In the school where I did my interviews, one mathematics teacher had a student who learned to solve a problem with help from an AI, but the AI taught her an incorrect method and she came to an incorrect answer. This caused her to fail her test (anonymous participant, personal communication, May 7, 2026).
By outsourcing cognitive labour to AI students may reduce their incentive to develop independent analytical skills, while simultaneously increasing the influence of large technology companies over their way of thinking. As students become more dependent on proprietary systems, these companies gain greater power in shaping how knowledge is produced, accessed, and understood.
Apart from students and tech companies, another power shift is visible between students and teachers. Teachers are losing power as homework is no longer an effective way to force students to engage with the material. They are searching for a way to gain their power back by adapting their assignments and assessments, as mentioned earlier. School policies have not yet caught up to the technological developments and teachers are waiting for guidance on how to integrate AI effectively. Currently, the only established policy is the EU AI Act (Future of Life Institute, 2026), and the schools I examined did not have any formal AI policies but rely on teachers to regulate student AI use autonomously.
While students benefit from the convenience of the ubiquity, availability and accessibility of unguided and unregulated AI in the short term, it can be harmful in the long term. Unrestricted AI use is particularly concerning as literacy is already declining; a third of high school students is functionally illiterate (Duin, 2026). Tech companies do benefit because more users means more data and they could start to benefit from possible dependencies. Teachers and institutions are negatively affected, as teachers struggle with the increased ease and frequency of AI-assisted cheating and schools and institutions are struggling to establish policies and norms surrounding AI use.
#
The current problem is that teachers and institutions are losing power and policies are not being developed quick enough to help them. However, currently all the power and responsibility are still placed on the teachers to individually adapt until there are regulations on how they can adapt in the most effective way. A substantial proportion of teachers across OECD countries report work-related stress, particularly related to workload and administrative demands (OECD, 2025).
My vision imagines that we redistribute power from teachers to students. Instead of focusing on how teachers can enforce AI regulations, which are currently often none existent or underdeveloped, we centre students in this vision, involve them in the discussion rather than decide for them, and empower them to make informed and ethical decision on their own AI use. According to Neil Sewyn, students should be involved into the discussion around generative AI, especially since the current conversations about cheating can also be regarded as a conversation about trust in students (DELab UW, 2025). Furthermore, Selwyn argues that AI literacy is critical for people who are living in an AI-driven world. He explains that AI literacy involves awareness of AI, which includes working knowledge on how AI is trained and awareness of where AI is, but also having the critical skills to work with the technology, work around it and work against it.
In my vision, we don’t approach students as passive consumers, but as active participants who have autonomy and are capable of making informed, ethical decisions and use AI tools responsibly. Rather than viewing the ubiquity, availability, and accessibility of AI as a problem, these characteristics can be used to students’ advantage by equipping them with the knowledge and skills to self-regulate their AI use responsibly. Since it is difficult to enforce the exclusive use of guided AI tools outside of school environments, a more effective approach is to teach students how to critically engage with AI tools themselves and use them ethically and responsibly. Ultimately, this vision reframes AI as an opportunity for student empowerment, where informed agency and critical digital literacy play a central role.
To materialise my vision, I created a structured lesson plan that translates my abstract vision into a tangible and actually usable artifact. I chose a lesson plan because it is a concrete educational tool that can be directly implemented inside the classroom. Take, for example, the new advice from the Onderwijsraad in which they advise to look differently at examinations (Onderwijsraad, 2026). They give an advice on what to do, but leave it open for schools and institutions on how to implement that. However, this means there is little guidance for schools as there is no concrete plan, and they do not have the time nor stress-levels to develop their own implementation plan. This is a pattern I have often noticed, particularly in news articles, where AI is criticised or future directions are proposed but without offering practical solutions that teachers can implement presently. I have created a lesson plan as it is the most practical and ready-for-us visualisation I can provide.
Lesson planning is one of the two most time-consuming tasks that teachers face, outside of giving the lessons (OECD, 2025). There are many AI tools currently developed intended to facilitate teachers by taking over the work, which adds a layer of irony in my opinion. It also showcases that AI really is becoming ubiquitous. I envisioned these lessons as part of the computer literacy course that many high schools offer, as I believe it fits in well with the module on Excel use. This means that the AI literacy course only takes place once a week for a limited number of weeks. Therefore, I have chosen to create lesson plans for four lessons to convey the vital information needed for a basic understanding in AI literacy. The topics are based on the research and interviews that I have done for this course, which showed me what the basic components of AI literacy are.
The first lesson focuses on defining and understanding GenAI, providing a foundational introduction necessary for the rest of the course. The second lesson focuses on using GenAI in a school context, including prompt creation and critical evaluation of outputs, ensuring the focus of this course remains on educational use, while the content is still applicable to other contexts. The third lesson addresses ethical AI use, including risks, responsibilities, and reflection on how AI use aligns with students’ values, as well as the risk of over-reliance and outsourcing cognitive work. The final lesson explores the AI production cycle, including hidden labour and environmental and social costs. This is important for helping students understand that AI is not neutral nor free, but depends on a significant number of material and human resources.
#
For me, the interviews were very beneficial. During the interviews, I realised that the students recognise themselves as to be not very knowledgeable on AI, even though they use it frequently. Student 1 regarded that looking up information is easier with GenAI than using the internet (meaning search engines) and she considered AI and the discussion surrounding it to be “not so serious”. Contrary, student 2 did seem slightly more knowledgeable and stated that they used GenAI sparingly for schoolwork and with quite some consideration. A noticeable difference between the two students was that student 2 had received something akin to a speed course on GenAI during their French lesson, which consisted of five students. This made me believe that basic AI literacy could make a significant impact on how students use AI.
The interviews also helped me realise just how ubiquitous, accessible and available GenAI is. Student 1 stated GenAI is “just there” and mentioned how you do not need to think about it to use it. Student 2 mentioned how you cannot escape it, as even Google forces a GenAI answer as the first result.
The students clearly had different experiences using GenAI. This made me realise I had been thinking of “high school students” as a collective noun but not as separate entities with individualities and autonomy. In the media, if they are involved into the conversation at all, students are often portrayed as one voice, or a few similar voices, but hardly as nuanced individuals with wildly differing perspectives. My vision was inspired by this to highlight the students as active participants in the discussion of GenAI and not passive consumers. I wanted to create a vision that highlighted their autonomy
Especially before the start of the course, I had an internalised, negative opinion of the use of GenAI in high school, not substantiated by well-invested arguments. I tried to reduce any personal bias by being highly aware of my question stating – making sure I did not ask any leading questions but solely open and unguided ones. I tried to ensure that the interviewees had enough time to share their experiences and continue down their own train of thoughts while still guiding them to keep the discussion relevant. If they remarked something I considered interesting, appropriate or surprising, I would ask them to “tell me more” about it. At the end of the first interview I had started thinking about AI literacy classes, which had not come up naturally. Therefore, I asked the students first if they had had any classes or explanations on GenAI, a slightly guided question, and subsequently what their thoughts are on AI literacy classes in the school. I believe this question did not prompt any confirmation bias, as I did not ask “do you think it would be useful/useless” but phrased it as an open question and because I got one positive and one neutral-negative response, where they did not think it was “handy”. However, not long after, they did mention unprompted that it might be useful for first years.
There were a few limitations to the interviews. Firstly, I have focussed my interview, and therefore my problem statement and vision on the older students, as that would limit any issues regarding consent. However, younger students are just as, if not more susceptible and capable of using GenAI. Another limitation is my memory bias. Based on the recommendation, I did not record the interviews but wrote down quick notes. I made sure to write about the interviews immediately after when it was still fresh in my mind, but it is quite possible that my recollection of the interviews is incomplete.
One limitation I had not expected, which was highlighted as I spoke to students first and afterwards to a few teachers, is that the students were not as ready to freely share their experiences and opinions without prompting. If I did not ask a question, they would patiently wait for me to ask the next one, and if they could answer a question with a single sentence, they would do so. I would try to prompt them to keep talking, but I also realised I might need some practice to become a good interviewer.
#
When I started this course, I was very hesitant to make myself heard in the discussion surrounding AI as I felt I did not know enough to be able to make a valid contribution. I had a stance akin to technological defeatism; I disliked the rise of AI in general and could not envision a future where we can successfully regulate AI and its developments. I knew high school students were frequently using GenAI and that teachers and institutions are struggling to update their policies to keep up with the technological developments. I believed that students would mostly use GenAI for writing essays and other assignments such as the profielwerkstuk (high school thesis).
With the discussion about technological defeatism and the examples presented in class, particularly the ones about radio and the anti-noise-pollution-campaign, I realised that technological defeatism is not a logical stance to take. I became more hopeful that we will find a way to regulate AI.
Because my views on AI were so negative, I wanted to challenge myself by creating a vision that did not remove AI from the equation, such as envisioning a way to redesign our education system where AI is no longer of influence, but a vision that would embrace AI with its current capabilities in a positive manner.
The interviews helped me regard AI even more in a positive light when they described how to they used GenAI when a teacher was failing them. For example, student 1 mentioned that their history teacher does not use a book, but relies solely on slides and if you have trouble learning from the slides, GenAI is a great way to ask for clarification on a topic or to create a mock exam based on the slides. This made me realise there are positives to AI that I have been ignoring to convince myself of the black-and-white “AI is bad” stance I had taken. I do still struggle with the idea that every solution I could create would be ignoring a different major problem. For example, I have now created a vision where students can use GenAI effectively, ethically and responsible but it ignores the cost and labour that is needed for AI (although I tried to include this particular problem somewhat by having one of the classes in my lesson plan on this subject).
I think the speculative intervention assignment made me try to think more creatively and made me more hopeful that in this vast world there must be someone who has a creative and brilliant idea to solve particular issues, as long as we encourage this kind of speculation and support and discuss people’s alternative visions.
Although I do now wonder if we will need to regulate AI in the future, as we also discussed in class how AI is currently a massive loss leader but it is not clear how they will start making a profit, and it is possible that in the future AI will no longer be so easily accessible as it is now.
Akyol, Ö. (2026, February 11). Alsof leerlingen die schrijfopdrachten in deze tijd van AI nog zelf gaan doen. BD.nl. https://www.ad.nl/binnenland/alsof-leerlingen-die-schrijfopdrachten-in-deze-tijd-van-ai-nog-zelf-gaan-doen~a022aff6/
Bardoel, F. (2026, April 16). Examenleerlingen op Koning Willem II College in Tilburg betrapt op geraffineerde spiekactie, expert slaat alarm. BD.nl. https://www.bd.nl/tilburg/examenleerlingen-op-koning-willem-ii-college-in-tilburg-betrapt-op-geraffineerde-spiekactie-expert-slaat-alarm~a72f62be6/
Basiouny, A. (2024, August 27). Without guardrails, generative AI can harm education. Knowledge at Wharton. https://knowledge.wharton.upenn.edu/article/without-guardrails-generative-ai-can-harm-education/
Bastani, H., Bastani, O., Sungu, A., Ge, H., Kabakcı, Ö., & Mariman, R. (2025). Generative AI without guardrails can harm learning: Evidence from high school. Proceedings of the National Academy of Sciences, 122(26). https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4895486
Boretti, A. (2026). Hallucinations in generative AI: A threat to scholarly integrity and the urgent need for publisher-led academically supervised verification. Energy Research & Social Science, 136, 104720. https://doi.org/10.1016/j.erss.2026.104720
De Kort, N. (2026, April 6). Docenten kijken na met AI: Slimme tijdsbesparing of risicovol? https://nos.nl/artikel/2609282-docenten-kijken-na-met-ai-slimme-tijdsbesparing-of-risicovol
DELab UW. (2025, sep 17). Neil Selwyn – Generative AI and Higher Education – Some Things We Need to Talk About . YouTube. https://www.youtube.com/watch?v=wZ699uPO5DA
Future of Life Institute. (2026). EU Artificial Intelligence Act | Up-to-date developments and analyses of the EU AI Act. https://artificialintelligenceact.eu/
Google. (n.d.). Find information in faster & easier ways with AI Overviews in Google Search—Computer. Google Search Help. https://support.google.com/websearch/answer/14901683
Laarhoven, J. van. (2026, March 3). Laat een kind zich verwonderen over AI maar ook verder kijken, is het advies aan ouders en docenten. ED.nl. https://www.ed.nl/helmond/laat-een-kind-zich-verwonderen-over-ai-maar-ook-verder-kijken-is-het-advies-aan-ouders-en-docenten~aa94d272/
Last, B., & Saçan, E. (2026, January 26). Staar je in het onderwijs niet blind op AI, het mag wel wat minder. bd.nl. https://www.ed.nl/opinie/staar-je-in-het-onderwijs-niet-blind-op-ai-het-mag-wel-wat-minder~a4f07721/
OECD (2025), Results from TALIS 2024: The State of Teaching. Paris: OECD Publishing. https://doi.org/10.1787/90df6235-en.
OECD (2025), Teaching and Learning International Survey (TALIS) 2024 Conceptual Framework, OECD Publishing, Paris, https://doi.org/10.1787/7b8f85d4-en.
Onderwijsraad. (2026, May 28). Kijk anders naar toetsing. Ministerie van Onderwijs, Cultuur en Wetenschap. https://www.onderwijsraad.nl/documenten/2026/05/28/kijk-anders-naar-toetsing
Thies, E. (2025, June 24). Maakt AI ons dommer of slimmer? ‘Het grote verschil zit hem in hoe je ChatGPT inzet.’ AD.nl. https://www.ad.nl/wetenschap/maakt-ai-ons-dommer-of-slimmer-het-grote-verschil-zit-hem-in-hoe-je-chatgpt-inzet~a6601b87/
Timmers, F. (2025, November 11). Twentse AI-expert waarschuwt: ‘Straks wordt school nog leuk ook!’ Tubantia. https://www.tubantia.nl/hengelo/twentse-ai-expert-waarschuwt-straks-wordt-school-nog-leuk-ook~a66ba6e6/
Van Duin, E. (2026, February 12). Maakt AI onze kinderen dommer? ‘Door de stroom aan prikkels uit onze schermen, stompen onze hersenen af.’ BD.nl. https://www.ad.nl/gezond/maakt-ai-onze-kinderen-dommer-door-de-stroom-aan-prikkels-uit-onze-schermen-stompen-onze-hersenen-af~aeb4781f/
Van der Valk, T. (2024, September 4). Kunstmatige intelligentie op de middelbare school: “Het is een enorm ding.” https://nos.nl/artikel/2535868-kunstmatige-intelligentie-op-de-middelbare-school-het-is-een-enorm-ding
Van der Vleuten, O. (2026, January 26). ChatGPT in de klas. Nel Verhoeven maakte een handleiding: ‘AI gebruiken is niet per se frauderen’ | Regio | AD.nl. https://www.ad.nl/vlissingen/chatgpt-in-de-klas-nel-verhoeven-maakte-een-handleiding-ai-gebruiken-is-niet-per-se-frauderen~a1014927/
Van Spronsen, L., & Halfhide, R. (2025, April 9). Middelbare scholen passen schoolopdrachten aan om AI-gebruik tegen te gaan. https://nos.nl/artikel/2562959-middelbare-scholen-passen-schoolopdrachten-aan-om-ai-gebruik-tegen-te-gaan
Vendrell, M., & Johnston, S.-K. (2026). Scaffolding critical thinking with generative AI: Design principles for integrating large language models in higher education. Computers and Education: Artificial Intelligence, 10, 100572. https://doi.org/10.1016/j.caeai.2026.100572
Westhreenen, S. van. (2025, June 17). Loop niet achter, omarm AI in het onderwijs. AD.nl. https://www.ad.nl/commentaar/loop-niet-achter-omarm-ai-in-het-onderwijs~afc0ac37/