Epson classroom survey: teachers ask for help supervising AI

A European survey about classroom expectations

TechRadar published its coverage of the Educate to Empower survey on 12 September 2026, according to TechRadar. The research puts teachers’ preparation and students’ expectations into the same discussion. Its useful contribution is to identify questions about supervision, rather than to establish what happens in every classroom where AI is used.

The countries surveyed were France, Italy, Germany, Spain, Poland and the UK, according to TechRadar. Print in the Channel describes a survey of 3,360 people conducted in July and August 2026, according to Print in the Channel. The geography and sample should stay attached to the findings; neither supplies an estimate for Algeria.

The survey was conducted by Focaldata and commissioned by Epson, according to Print in the Channel. That commissioning relationship matters when describing the evidence. The coverage presents responses to a survey, not an experimental test of learning outcomes, and publication by several outlets does not create several independent studies.

Supervision and personal use are separate training needs

In the survey, 68% of teachers believed AI use in homework negatively affected learning, according to TechRadar. The word ‘believed’ is essential: the result reports teachers’ perceptions. It does not establish that AI caused learning harm, quantify such harm or resolve how different uses might be assessed.

Among surveyed students, 76% expected to use AI for learning; of those students, 87% already used some form of AI for school at least weekly, according to Print in the Channel. The second figure concerns students who expected to use AI, not all students. Keeping that denominator explicit prevents a broader claim than the research supports.

The survey found that 82% of teachers wanted more training in supervising students’ AI use, according to mínuszos.hu. This request points to a specific professional-development question: how would a teacher examine a student’s process and decide what kind of guidance to provide? That is an evaluation question, not a claimed result of training.

Another 78% wanted more training and guidance on using AI in their own work, according to mínuszos.hu. Supervision of learners and teachers’ own use are distinct needs. A proposed training plan could state which of these it addresses, rather than assuming that familiarity with a tool automatically prepares someone to assess a learner’s use of it.

Assess the evidence before adopting it

For a prospective Algerian user, the first useful step is to define the decision this evidence could support. A training provider might want to revise an exercise; an employer might want to assess a particular task; an education team might want to examine supervision. These are proposed uses of the information, not findings about Algeria. Begin with the actual work and identify what a successful learner would have to demonstrate. That makes it possible to ask focused questions without treating an international announcement or survey as a local needs assessment.

Next, distinguish access, participation, confidence and demonstrated capability in the evaluation plan. A person can describe an intention without having carried it out, and an organization can announce a provision without establishing its effects. Ask what evidence would justify the next decision: a reviewed work sample, an observed task, or a documented explanation of a choice. These are suggested assessment methods rather than reported results. They give a local team a way to test relevance while keeping the published findings within their original boundaries.

Before committing resources, record what remains unknown for the intended setting. Consider the language of instruction, the tools an exercise would require, the time available to participants and who would review the work. None of those local conditions is established by the evidence presented here. A small, clearly defined evaluation could help an organization answer its own questions; it should have an explicit learning objective and a review point. Expansion should depend on what that evaluation shows, rather than on the prominence of the publisher or the size of a headline percentage.

Finally, keep the source of each conclusion visible when sharing the findings internally. Label the published evidence separately from the team’s proposed response, and state which population or product the evidence concerns. If a decision requires a local baseline, collect that baseline instead of borrowing an overseas percentage. This discipline does not require dismissing the announcement or survey. It means using it to formulate a question that can be answered in the intended workplace or classroom, with responsibility for interpreting the answer assigned before any broader adoption.

An honest comparison for Algerian training providers

For an Algerian teacher-training provider, the connection is a set of questions about supervision and foundational skills. The survey does not establish local demand, local usage or the effects of AI on Algerian students. A provider could ask educators to describe the guidance they would need for an actual assignment and how they would assess the student’s reasoning. It could separately examine the educator’s own work. That would be a proposed local inquiry, grounded in a clear distinction between supervising a learner and using a tool, without treating European perceptions as causal evidence or local statistics.

Frequently Asked Questions

Sources & Further Reading

• TechRadar — Teacher perceptions and surveyed countries
• Print in the Channel — Survey design and student responses
• mínuszos.hu — Teacher training requests
• 80 per cent of teachers worry AI is entering the classroom faster than schools can manage — Epson