Last updated: August 2026
EdTech is the set of technologies applied to education, from early childhood schooling to adult professional training. 2026 is the year generative AI stopped being an isolated experiment and became part of ministerial guidelines, while Italy simultaneously introduced a smartphone ban across all upper-secondary schools: two seemingly opposite moves coexisting within the same education policy.
Contents
- What edtech is
- AI tutors: the promise of one-on-one for everyone
- Adaptive learning: how it works
- VR and AR in the classroom: impossible labs
- AI and homework: the problem nobody has solved
- Italy: the smartphone ban and AI rules
- Micro-learning and certifications: learning as an adult
- Languages with AI: the end of traditional courses?
- The augmented teacher
- FAQ and school in 2035
What edtech is (definition)
EdTech (education technology) is the application of digital technologies to teaching and learning, across three distinct areas: K-12 (education from early years to upper secondary), higher education, and continuing professional training for adults. Each area follows different logic: K-12 favours tools that support the teacher and the classroom, higher education is seeing growth in online courses and AI-assisted research tools, and professional training is dominated by short certifications and micro-learning tied to labour-market skill needs.
AI tutors: the promise of one-on-one for everyone
AI-based tutors, such as Khan Academy’s Khanmigo or Squirrel AI, promise to replicate for every student the individual attention of a private tutor, a model pedagogical research has long associated with stronger learning outcomes. The typical approach is Socratic: instead of giving the answer, the tutor asks guiding questions and suggests intermediate steps, pushing the student to reason rather than copy.
On the evidence side, though, it’s honest to say public, shared metrics on the real cognitive impact of these tools are still missing: European reference frameworks, from DigCompEdu to the Joint Research Centre’s recommendations, consider assessing the cognitive impact of AI tutors an open field. Khan Academy itself, between October 2025 and April 2026, ran a series of internal tests on Khanmigo tracking mainly response latency and next-item correctness after a tutoring session, not third-party-validated learning gains. In Italy, the first ministerial trial of an AI tutor, run across 15 pilot classes in four regions over two years, concluded in May 2026: official results processed by INVALSI are expected in September 2026.
Adaptive learning: how it works
Adaptive learning is the technique by which a platform automatically calibrates exercise difficulty based on a student’s previous answers: after a run of correct answers the system gradually raises the difficulty; if a student repeatedly gets a certain type of exercise wrong, the system steps back and offers reinforcement content on that specific concept before moving on. It’s the same principle behind spaced-repetition vocabulary apps, now applied to broader subjects such as maths or reading.
VR and AR in the classroom: impossible labs
Virtual and augmented reality let schools simulate experiences that would otherwise be impossible or too costly in a real classroom: a virtual dissection instead of a chemistry lab with hazardous reagents, an immersive reconstruction of a historical site instead of a textbook page. The practical limits remain twofold: headset costs for a whole class, still significant for many Italian schools, and the need for specific teacher training, since teachers often lack the time to properly integrate the tool into their lesson planning. Italian cases today remain isolated pilot projects, mostly funded through the PNRR recovery plan, rather than a standard school-wide feature.
AI and homework: the problem nobody has solved
The hottest topic for teachers and families remains homework done with the help of, or entirely by, artificial intelligence, and the technological answer, AI-generated text detectors, has proven structurally unreliable. A 2023 Stanford University study that tested seven commercial detectors on essays written by non-native English speakers recorded a false-positive rate above 60%: texts written entirely by people, but flagged as AI-generated, precisely because of the more regular, controlled style typical of people writing in a second language. A more recent analysis found that these tools’ accuracy collapses from 39.5% to 17.4% when an AI-generated text is even lightly reworded. It is no coincidence that OpenAI retired its own AI text classifier back in 2023, acknowledging its accuracy was insufficient.
The practical consequence for schools is that no detector can today be used as definitive proof against a student: at most it can be a lead to verify through other means (an oral interview, comparison with earlier work), never an automatic verdict. The debate is therefore shifting from a technological arms race to rethinking assessment itself, with growing weight given to oral exams and supervised in-class work.
Italy: the smartphone ban and AI rules
Two separate regulatory tracks moved in parallel in Italy between 2025 and 2026. On smartphones, the ban already in force for years in early years, primary and middle schools was extended by Education Minister Giuseppe Valditara to upper-secondary schools too, starting with the 2025/2026 school year, through Circular no. 3392: students must hand in their phone at the entrance or keep it switched off and inaccessible for the entire school day, with exceptions for students with disabilities under an Individualised Education Plan or for specific IT and telecommunications technical tracks. In February 2026 the Ministry launched a national monitoring exercise on how the circular was being applied: by March 2026, 90% of secondary schools had already written the new rules into their internal regulations, with 64.2% of schools requiring the phone to be kept switched off in the student’s bag, 25.5% using dedicated in-class storage boxes, and 5.5% requiring mandatory hand-in at the entrance.
On the opposite front, artificial intelligence, the Ministry of Education and Merit published its Guidelines for the Introduction of Artificial Intelligence in Educational Institutions on 9 August 2025, attached to Ministerial Decree no. 166/2025, asking schools to adopt a human-centred, ethical approach, to choose certified and secure AI system providers, to protect minors’ data, and to ensure fairness and transparency. The regulatory framework was then reinforced by Law no. 132 of 23 September 2025, the first national law in Europe entirely dedicated to artificial intelligence, in force since 10 October 2025. The obligation for school staff to receive AI-literacy training, required under the EU AI Act, has formally applied since 2 February 2025, but its practical rollout in Italian schools starts on 3 August 2026: in the meantime, over 5,760 training courses on using AI in teaching have already been delivered using PNRR recovery funds.
Micro-learning and certifications: learning as an adult
Adult training has seen the rise of skill-based hiring, recruitment based on verifiable competencies rather than traditional qualifications, which has fuelled growth in short courses and certifications focused on a single technical skill. It’s worth being honest with readers on one point: the value of these certifications on a CV varies enormously depending on who issues them, a technical certification recognised by a major industry vendor carries very different weight from a generic completion certificate from an online course platform, and recruiters do not treat them the same way.
Languages with AI: the end of traditional courses?
AI conversation apps let learners practise speaking a foreign language at any time, without needing an available human partner. The real results, though, remain more modest than the marketing around them: AI is effective for practising fluency and vocabulary in a low-stakes setting, but it does not yet fully replace interaction with a native speaker for cultural and pragmatic nuance, nor structured grammar teaching within a formal learning path.
The augmented teacher
This pillar’s thesis is that artificial intelligence, in its most mature school use, does not replace the teacher but shifts their time: fewer hours spent on mechanically correcting repetitive exercises, more time devoted to lesson design, individual support for struggling students, and managing the classroom as a social space. It’s a recognisable, checkable position, not a vague forecast: the MIM’s own guidelines insist that AI must support teachers’ work without replacing it, since the school remains a space for direct human relationships that no digital tool can replicate.
FAQ
Is artificial intelligence banned in Italian schools?
No: the MIM guidelines (DM 166/2025) regulate its responsible introduction, they do not ban it. What is banned is smartphone use throughout the school day, in all schools from early years to upper secondary, with specific exceptions.
Does VR damage children’s eyesight?
Prolonged VR headset use is not recommended for younger children according to manufacturers’ own guidelines (often advised against below a certain age), but moderate, supervised classroom use is not the same as prolonged, unsupervised recreational use.
Are AI text detectors reliable for catching a copied assignment?
No, not definitively: independent studies show false-positive rates above 60% in some contexts, which is why they should be used at most as a lead to verify, never as sole proof.
Do AI tutors really work better than a textbook?
Early evidence is promising on engagement and personalisation, but independent, published studies proving validated learning gains at scale are still missing; results from Italy’s ministerial trial are expected from INVALSI in September 2026.
Are online certifications worth as much as a degree?
It depends on the field and issuer: in some specific technical areas they can carry as much or more weight in hiring decisions, but they are not equivalent to a degree in terms of overall legal and educational recognition.
School in 2035
If the convergence between AI rules and restrictions on personal devices keeps maturing, the school of the next decade could look like an environment more controlled in its access to consumer technology, but richer in targeted teaching tools: AI tutors integrated into everyday teaching, oral assessment brought back to the centre of the system, and a teacher’s role increasingly oriented towards lesson design and individual support rather than frontal delivery of content. The open question is whether the Italian school system will have the resources, in training and infrastructure, to keep pace with the rules governing this change.
Sources
- Italian Ministry of Education and Merit (MIM), Circular no. 3392 and Ministerial Decree no. 166/2025 (AI guidelines for schools)
- Law no. 132 of 23 September 2025, Italy’s first AI law
- Khan Academy, internal data on Khanmigo improvement tests (October 2025-April 2026)
- Stanford University, study on false-positive rates of AI text detectors on TOEFL essays
- Agenda Digitale, analysis of Italy’s ministerial AI-tutor trial
- Save the Children, Sky TG24, coverage of the smartphone ban extension to upper-secondary schools
