A student receives a zero because the school’s detector says her essay was written with artificial intelligence. Her family objects, the teacher has doubts, and the story fits neatly into a headline. The decisive detail is not that a new tool exists or that a conflict has occurred. It is that nobody can explain what evidence the detector produced, what error rate it allows or how the student can appeal. At that moment, an individual incident reveals a public problem: an institution is making decisions that affect rights and opportunities through a procedure that may not be understandable, reviewable or equal for everyone.
The detail is not decoration: it changes the question#
News about school technology is often presented as a stream of novelties: tablets arrive, phones are banned, a platform is introduced, data is leaked or a student uses a chatbot. Yet novelty alone is not a public problem. To recognise one, we need to identify the mechanism connecting the case to a collective obligation. It might be a rule without clear criteria, a purchase that shapes teaching for years, an economic barrier, excessive data collection or the absence of an accessible alternative. A useful detail lets us move from “this happened to someone” to “this could predictably happen to other people because the system is organised this way.”
News about school technology is often presented as a stream of novelties: tablets arrive, phones are banned, a platform is introduced, data is leaked or a student uses a chatbot.
Consider three headlines. “A pupil forgets a password and misses a lesson” describes an inconvenience. “The password recovery process requires a personal smartphone” reveals a condition that excludes those without one. “A teacher takes twice as long to prepare a digital activity” may be an individual difficulty; if the authority introduced the platform without training, time or support, the fact points to an incomplete policy. “A family rejects an application” is an opinion; if there is no information about the data processed and no alternative way to participate, it becomes a question of transparency and access. The detail changes the interpretation because it identifies a rule, a distribution of resources or an institutional choice.
The finding should be expressed in a testable sentence: “To submit homework, every learner needs an internet connection outside school hours, and the school provides neither loans nor an alternative space.” That is more useful than saying “digitalisation increases inequality.” The broader statement may be true in some settings, but it is too wide to guide action. The first identifies who must do what, allows the school to count how many people are excluded and opens concrete solutions. UNESCO’s 2023 GEM Report similarly argues that education technology should be judged by its appropriateness to context, equity, evidence and sustainability, rather than by its mere presence.
Four tests for recognising a collective problem#
The first test is repetition. We do not need to wait for a hundred victims, but we should look for signs that the case arises from a stable condition. Does the same rule apply to an entire year group? Do other teachers encounter the same obstacle? Does the incident return every week? Recording dates, affected groups and responses received is more useful than collecting disorderly screenshots. A simple table — event, consequence, relevant rule, school response — helps distinguish an isolated failure from a pattern.
The second test is the unequal distribution of harm. An apparently neutral measure may require resources that not every family has. If an activity needs a printer, mobile data, a recent device, a quiet home or adult help, the burden is not shared equally. There are less visible inequalities too: an interface may be incompatible with screen readers; a video without captions excludes; complex authentication makes access harder for some pupils; and automated translation may distort communication with newly arrived families. Asking “Who will find this rule hardest to meet?” often uncovers the public significance of the case.
The third test is the importance of the consequence. Not every inconvenience needs a campaign. We should examine whether it affects assessment, attendance, privacy, participation, wellbeing, curriculum access or the ability to appeal. A web filter blocking a minor page for five minutes does not carry the same weight as a system preventing an examination from being submitted. An error in a reading recommendation differs from an automated classification that directs a learner towards a lower-expectation pathway. The harder a consequence is to reverse, the stronger the human oversight and explanation should be.
The fourth test is institutional capacity to prevent or repair. A problem becomes public when the solution cannot rest only on the pupil “adapting” or the family buying something. The school, authority or supplier controls the design, contract, rule or appeal channel. This does not mean finding someone to blame immediately. It means locating the level at which a reasonable adjustment — a non-digital alternative, human review, teacher training, a limited pilot or a contractual clause — could prevent recurrence.
How to investigate without turning suspicion into accusation#
The first step is to reconstruct the process. Ask what happened before, during and after: what task was set, which tool was used, what instructions learners received, what data triggered the decision and who confirmed it. Facts, interpretations and unanswered questions should be separated. “The platform recorded the submission at 00:03” is verifiable; “the system wanted to punish him” assigns an intention that nobody may be able to prove. Precision protects the affected person and makes it harder for the conversation to drift into an emotional dispute.
The second step is to find the current rule and its stated purpose. This may be a teaching plan, school regulation, family notice, assessment protocol or tool terms. “We have always done it this way” is not sufficient. Ask which educational objective the measure pursues and how the school knows it achieves that objective. If the aim is to limit distraction, there may be less intrusive options than monitoring every action on a device. If the aim is to detect plagiarism, schools need pedagogical evidence and a checking procedure, not an automated score treated as a verdict.
The third step is to hear different perspectives without turning the meeting into a referendum of impressions. The learner can describe the actual journey; the teacher can explain workload and educational purpose; the family can describe home conditions; technical staff can identify system limitations; and leadership can clarify organisational duties. One productive question is: “What evidence would make us change our mind?” It forces everyone to state what is missing. Learners should also be asked what works, what fails and which remedy would be practical. UNESCO’s youth report on education technology calls for choices that are appropriate, equitable, evidence-based and sustainable, with learners at the centre.
The fourth step is to compare claims with independent sources. A sales brochure describes functions; it does not demonstrate educational impact. A supplier’s survey may be informative, but readers need to know who responded and how. Strong claims — “personalises learning,” “detects cheating,” “improves attention” — require observable measures and comparison. If specific evidence is weak, the responsible conclusion is not that technology is useless. It is that the school should test it on a small scale, define indicators and retain a reversible exit.
A conversation script for families and schools#
A family can begin without accusing: “We want to understand how this decision was made and prevent the problem recurring.” It can then ask five questions: What was the educational objective? Which rule or criterion was applied? What evidence was reviewed? Who can reconsider the outcome? What alternative exists for someone who cannot or does not wish to use the tool under the same conditions? A written response is reasonable where assessment, data or access is affected, but the message should be brief and concrete. A half-page chronology is usually more effective than twenty forwarded messages.
The school can respond better by avoiding two extremes: denying the issue to protect its reputation, or suspending every innovation out of fear. A mature response states what is known, what still needs checking and when review will finish. It can appoint a responsible person, preserve the original work, temporarily suspend a hard-to-reverse consequence and offer an equivalent route. If several cases exist, the change should be communicated to the whole community without identifying the person who raised the alarm.
For a solution to be assessable, it needs an owner, a date and an indicator. “We will improve communication” is not enough. “Before next term, every compulsory tool will have a sheet explaining purpose, data processed, support, alternative and incident channel” can be checked. “We will train teachers” is vague; “two practical sessions and supported use during the pilot, with resolved problems recorded” creates learning. A school need not promise perfection, but it should show that it notices consequences and corrects them.
From a concrete case to a prudent school policy#
Good technology policy starts with an educational need, not a catalogue. Before acquiring or mandating a tool, the team should describe the problem it aims to solve, non-technological options, people who might be excluded and the cost of leaving. It can then run a limited pilot, gather experiences from different groups and decide whether to expand, modify or stop. The question “What will we stop doing when we choose this?” brings teacher time, attention, maintenance and supplier dependence into view.
Learner participation does not mean letting pupils decide everything; it means recognising their knowledge of actual use. A student council may notice that notifications interrupt concentration, two platforms duplicate assignments or an accessibility feature fails. UNESCO’s AI competency framework for students proposes progression from understanding to applying and creating, grounded in a human-centred and ethical perspective. That approach encourages schools to teach how systems work and fail, how outputs are verified and when refusal is legitimate, rather than merely allowing or banning applications.
An inventory is also essential: active tools, purpose, classes, data, responsible person, review date, cost and alternative. Without an inventory, every incident looks isolated. With one, a school may discover that five applications request the same information, a licence is barely used or nobody knows who handles complaints. Publishing an understandable version builds trust and lets families ask questions before harm occurs.
Warning signs and proportional responses#
Some signs justify pausing deployment: automated decisions without human review; criteria that cannot be explained; compulsory personal accounts; exposure of sensitive information; no accessible alternative; penalties for technical failures; disproportionate monitoring; or a contract that makes it difficult to retrieve work and data. A pause is not a permanent condemnation. It creates time to clarify purpose, risks, responsibility and repair.
Other difficulties can be fixed without drama: a confusing interface, insufficient training, excessive notifications or contradictory instructions. A short guide, nearby support and an adjustment period may be enough. Proportionality links response to harm. Not every news story deserves to become a crusade, but no case should be dismissed as an anecdote before checking whether it reveals a rule affecting many people.
A case sheet that turns news into public learning#
The community can document each case on one page with eight fields: observed event; people affected; consequence; condition that enabled it; available evidence; unanswered questions; the decision-maker able to change it; and a proposed measure with a review date. It should also record what alternative was offered and how someone can participate without public exposure. This sheet depersonalises conflict and creates institutional memory. When several sheets are reviewed together, patterns emerge that no isolated story could show.
The decisive detail will not always be dramatic. It may be a password available only by phone, an assignment incompatible with assistive technology, a penalty nobody can appeal or a family buying a device for the third time. Examining it rigorously does not mean resisting change. It means requiring change to serve an educational purpose, distribute benefits and burdens fairly, allow review and preserve room for human judgement. The news story then moves beyond fleeting outrage: it becomes a well-formed public question and, ultimately, an improvement that can be verified.
