Skip to content
ETHORITY®
Search ETHORITY
DE

Global edition · EN

36% of lower-secondary teachers used AI at work (OECD).

From 2 Dec 2027 high-risk duties apply to certain AI in admission, grading and proctoring. Where AI already helps schools, universities and corporate learning.

v1.4 · 40 pages · 27 Oct 2026 · Edition

Research populations, geographical scope, source dates and limitations are stated in this edition. The research catalogue and the interactive website catalogue cover different populations.

Choose a licence for this edition →

Preview: pages 1–5. Full PDF with a paid licence.

36% of lower-secondary teachers used AI at work (OECD). · EN · v1.4
Global edition · EN · v1.4

Look inside this edition

36% of lower-secondary teachers used AI at work (OECD). · Page 2 · EN
Page 2 · Page 2/40
36% of lower-secondary teachers used AI at work (OECD). · Page 3 · EN
Page 3 · Page 3/40
36% of lower-secondary teachers used AI at work (OECD). · Page 4 · EN
Page 4 · Page 4/40
36% of lower-secondary teachers used AI at work (OECD). · Page 5 · EN
Page 5 · Page 5/40
Edition, source scope and file

Good_AI_Prism_Sector_Brief_Education_EN_v1.4_20261027.pdf · Edition 1.4

Research populations, geographical scope, source dates and limitations are stated in this edition. The research catalogue and the interactive website catalogue cover different populations.

Research method and catalogue coverage →
1 in 3
lower-secondary teachers used AI in their work across OECD systemsOECD TALIS 2024 · oecd.org ↗
24
AI offerings mapped in EducationGood AI Prism, Oct 2026
2 Dec 2027
EU high-risk duties for AI in admission, grading and exam proctoring applyEU AI Act (as amended by Reg. (EU) 2026/1744) · eur-lex.europa.eu ↗
44m

The world is 44 million teachers short.

The world needs 44 million more teachers by 2030, so AI arrives in a profession already under pressure.

UNESCO and Teacher Task Force, Global Report on Teachers

7 of 24

Corporate learning leads the market map.

Seven of the 24 offerings mapped in Education serve corporate learning and course creation.

Good AI Prism 2026/27 study dataset, Oct 2026

+0.31 SD

Guided tutors add 1.5 to 2 years of learning.

Six weeks of guided GPT-4 tutoring in Nigeria raised learning by 0.31 standard deviations, about 1.5 to 2 years of schooling.

World Bank, De Simone et al., 2025

What's inside

  1. Students first, teachers fast behind. Who uses AI in schools, universities and academies, and why design decides the effect.
  2. AI arrives in a profession under pressure. Teacher shortages, skills gaps and what teachers expect from AI.
  3. Grading and proctoring: high-risk duties from Dec 2027. The AI Act dates, Germany's building blocks and the rules in nine jurisdictions.
  4. Half of the 24 offerings help teachers or tutor students. The market map by job group and headquarters.
  5. Choose tools built for learning. Seven questions before a pilot, a workforce annex and a 90-day start.

The brief and the studies behind it

About this brief

Who is this brief for?

For leaders of school authorities and school groups, universities, vocational schools, corporate academies, education ministries and EdTech providers who need to decide which AI belongs in learning and teaching.

What does the Good AI Prism map?

The Good AI Prism by Sten Franke & ETHORITY® maps the AI market on one wheel of 24 topic areas. In Education it places 24 offerings, each recorded with a dated source from the provider's own pages, and groups them by job: corporate learning, teacher assistants, tutoring, study tools and assessment.

Which edition is this, and where are earlier ones?

This is the Global edition in English, version 1.4 of 27 Oct 2026. Editorial revision: openings, claim scope and legal wording; navigation to ethority.ai added. v1.0 appeared on 22 Oct 2026, and the imprint lists every edition since. Earlier editions remain available, and a German edition for Germany appears alongside.

How do I cite it?

ETHORITY (2026). Good AI Prism Sector Brief: Education. Global edition, v1.4, October 2026. https://ethority.ai/research/publications/education-global-en

ETHORITY · LONG READ

Six weeks of guided AI tutoring, 1.5 to 2 years of learning.

One in three teachers across OECD systems already uses AI, and students are further ahead. The Education brief shows what lifts learning, what the market offers and which rules apply from 2 Dec 2027.

Design decides the effect

Six weeks of guided GPT-4 tutoring in Nigeria raised learning by 0.31 standard deviations, about 1.5 to 2 years of typical schooling, a World Bank study found. The same technology without guardrails did the opposite: Turkish students who practised with plain GPT-4 scored 17% worse on a later exam, according to a study in PNAS. The OECD Digital Education Outlook 2026 reaches the same conclusion: tools built for teaching work better than general chatbots.

That is the central finding of the Good AI Prism Sector Brief: Education. AI is already in every classroom through students' phones. The decision in front of schools, universities and company academies is which AI, designed how, and owned by whom.

Students first, teachers fast behind

Around one in three lower-secondary teachers used AI in their work across OECD systems in 2024, TALIS 2024 reports, from about three in four in Singapore and the UAE to 14% in France. Teachers use it most to learn about a topic (68%) and to plan lessons (64%); only a quarter apply it to student performance data.

Germany moved within a year: the share of teachers using AI for work several times a month or more rose from under a third to 52% (Deutsches Schulbarometer 2026). Students are further ahead. 84% of German 12- to 19-year-olds have used ChatGPT and 74% use AI for homework (JIM 2025). 95% of UK undergraduates use AI, 94% for assessed work (HEPI 2026), and 54% of US teens have used chatbots for schoolwork (Pew Research Center).

A profession under pressure

The world needs 44 million more teachers by 2030 (UNESCO and the Teacher Task Force). In Germany the education ministers project a gap of about 49,000 teachers to 2035, concentrated in lower secondary and vocational schools (KMK, Feb 2025). Skills are the main barrier: three in four teachers who do not use AI cite a lack of knowledge, and 52% of German teachers say they need training. Teachers are cautious, not hostile: two thirds expect negative effects on critical thinking, while 59% see benefits for individual support (Schulbarometer 2026).

The rules: high-risk duties for grading from Dec 2027

Emotion recognition in education institutions has been banned in the EU since 2 Feb 2025. From 2 Dec 2027, AI used for admission, grading, steering learning and exam proctoring carries high-risk duties under Annex III of the AI Act, with dates as amended by Regulation (EU) 2026/1744. Germany adds its own building blocks: the education ministers' recommendation on AI of October 2024, €5 billion for DigitalPakt 2.0 from 1 Sep 2026, and AIS.chat, a shared chatbot for all 16 Länder, hosted in the EU, with data not used for training (FWU). In England, the exam regulator asked exam boards to strengthen detection of AI misuse (Ofqual, Mar 2026).

What the market offers

The Good AI Prism places 24 offerings in Education. Seven serve corporate learning and course creation, among them Docebo Harmony, 360Learning AI Content Builder and Sana Learn. Six help teachers prepare lessons, such as MagicSchool, Brisk Teaching and fobizz. Six tutor students directly, including Khanmigo and AIS.chat. Three are course, language and study tools, such as Duolingo Max and NotebookLM, and two support assessment: Gradescope and Kahoot! AI tools. 12 of the 24 offerings come from providers headquartered in Europe, five of them in Germany. The Education area of the Prism shows every entry with its source and date.

What to do Monday

  1. List every AI tool in use, including the ones teachers and students bring themselves.
  2. Separate preparation and tutoring from grading, admission and proctoring, and map each use to the rules; the second group carries high-risk duties from 2 Dec 2027.
  3. Prefer tools built for learning, with pedagogical guardrails and evidence, over general chatbots.
  4. Train teachers first and document AI literacy measures, with a named teacher accountable for every grade.
  5. Pilot one guided tutoring or preparation use with a baseline, and scale only on measured learning gains, teacher time and equity.
Sources and further reading

YOUR NEXT STEP

Discuss your question with Sten Franke.

Bring your business question or the figures you already know. ETHORITY investigates within the agreed scope; together with Sten Franke you agree the next piece of work.

Send an enquiry →
36% of lower-secondary teachers used AI at work (OECD).

THE FINDING AT A GLANCE

36% of lower-secondary teachers used AI at work (OECD).

From 2 Dec 2027 high-risk duties apply to certain AI in admission, grading and proctoring. Where AI already helps schools, universities and corporate learning.

Read the sources and scope on the figure. The graphic is a separate format of this publication; it does not change the study’s research population.

Use this research in your work

Choose the readers and rights for this exact edition.

Global edition · EN · v1.4

€99 net · one-time

Enquire about this licence →
Need several studies? Compare annual plans →
PDF · SHA-256 a433ffc3d2045aeadacdc52c5be7600e918bd3b846d06ea614ad21c872c68907
Authorship, rights and AI assistance

Sten Franke & ETHORITY

© 2026 ETHORITY & STEN.AI LLC. Text and exhibits may be quoted with source attribution. Good AI Prism artwork: CC BY-NC-ND 4.0 (artwork only; data, software and third-party marks not covered); commercial use by licence on request: https://ethority.ai/products/research-licensing

Produced with AI assistance: research compilation, drafting, German and English language work, layout and rendering were carried out with AI systems (OSIRIS) under ETHORITY editorial direction (Sten Franke). Figures carry their cited sources.

Metadata identifies this edition. The SHA-256 identifies its exact file bytes.

Explore this research area

Briefings and background information