> **来源：[研报客](https://pc.yanbaoke.cn)** # OECD Digital Education Outlook 2026 Exploring Effective Uses of Generative AI in Education # OECD Digital Education Outlook 2026 EXPLORING EFFECTIVE USES OF GENERATIVE AI IN EDUCATION This work is published under the responsibility of the Secretary-General of the OECD. The opinions expressed and arguments employed herein do not necessarily reflect the official views of the Member countries of the OECD. This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. The statistical data for Israel are supplied by and under the responsibility of the relevant Israeli authorities. The use of such data by the OECD is without prejudice to the status of the Golan Heights, East Jerusalem and Israeli settlements in the West Bank under the terms of international law. # Note by the Republic of Türkiye The information in this document with reference to "Cyprus" relates to the southern part of the Island. There is no single authority representing both Turkish and Greek Cypriot people on the Island. Türkiye recognises the Turkish Republic of Northern Cyprus (TRNC). Until a lasting and equitable solution is found within the context of the United Nations, Türkiye shall preserve its position concerning the "Cyprus issue". Note by all the European Union Member States of the OECD and the European Union The Republic of Cyprus is recognised by all members of the United Nations with the exception of Türkiye. The information in this document relates to the area under the effective control of the Government of the Republic of Cyprus. # Please cite this publication as: OECD (2026), OECD Digital Education Outlook 2026: Exploring Effective Uses of Generative AI in Education, OECD Publishing, Paris, https://doi.org/10.1787/062a7394-en. ISBN 978-92-64-74128-7 (print) ISBN 978-92-64-91530-5 (PDF) ISBN 978-92-64-51513-0 (HTML) OECD Digital Education Outlook ISSN 2788-8568 (print) ISSN 2788-8576 (online) Photo credits: Cover © Gerhard Richter 2025 (23122025). Corrigenda to OECD publications may be found at: https://www.oecd.org/en/publications/support/corrigenda.html. $\odot$ OECD 2026 # Attribution 4.0 International (CC BY 4.0) This work is made available under the Creative Commons Attribution 4.0 International licence. By using this work, you accept to be bound by the terms of this licence (https://creativecommons.org/licenses/by/4.0/). Attribution - you must cite the work. Translations - you must cite the original work, identify changes to the original and add the following text: In the event of any discrepancy between the original work and the translation, only the text of the original work should be considered valid. Adaptations - you must cite the original work and add the following text: This is an adaptation of an original work by the OECD. The opinions expressed and arguments employed in this adaptation should not be reported as representing the official views of the OECD or of its Member countries. Third-party material – the licence does not apply to third-party material in the work. If using such material, you are responsible for obtaining permission from the third party and for any claims of infringement. You must not use the OECD logo, visual identity or cover image without express permission or suggest the OECD endorses your use of the work. Any dispute arising under this licence shall be settled by arbitration in accordance with the Permanent Court of Arbitration (PCA) Arbitration Rules 2012. The seat of arbitration shall be Paris (France). The number of arbitrators shall be one. # Editorial The OECD Digital Education Outlook is the OECD's flagship publication presenting our latest analysis of emerging digital technologies in education. This 2026 edition synthesises evidence and expert insights to show how generative AI has the potential to transform the quality and effectiveness of learning, as well as the productivity of education systems, provided its associated risks are carefully managed. Its applications include enhancing student learning, supporting teachers' performance while preserving professional autonomy, and strengthening education systems, as well as institutional and research capacities. For students, generative AI can scale personalised learning through intelligent tutoring systems, including in lowinfrastructure settings. Generative AI can also support knowledge acquisition by enabling collaborative learning and enhancing creativity. However, evidence shows that overreliance on generative AI tools that provide direct answers can reduce students' active engagement, improving task performance without corresponding learning gains. When used as a shortcut rather than a learning tool, generative AI can displace cognitive effort and weaken the skills that underpin deep learning. For example, a field experiment in Türkiye found that while access to GPT-4 improved short-term performance – by $48\%$ with the standard interface, and by $127\%$ with a tutoring version designed to support learning – students performed $17\%$ worse once access was removed, showing that generative AI can undermine learning unless explicitly designed to support skill acquisition. For teachers, generative AI can improve both productivity and teaching quality. Evidence cited in the report shows a $31\%$ reduction in time spent on lesson and resource planning by secondary science teachers in England, and a 9-percentage-point increase in student pass rates when low-experience tutors used AI support, with smaller gains for more experienced tutors. According to the OECD's 2024 Teaching and Learning International Survey, $37\%$ of teachers already use generative AI for work-related tasks - such as learning about or summarising topics and supporting lesson planning - with substantial variation across countries. At the same time, concerns persist that overreliance on AI could undermine teacher autonomy and professionalism, raise ethical risks, and, when used extensively for tasks such as marking, feedback or lesson planning, erode teachers' professional skills. The report calls for a shift towards educational generative AI systems designed with teachers, enabling them to monitor students' interactions with generative AI and actively shape its use in learning. At the system level, generative AI can improve the efficiency of education systems and school management by automating and supporting administrative and analytical processes. It can help develop standardised assessment items, review curricular alignment by analysing actual versus expected student workload, enhance study and career guidance, and support the classification of educational resources to name just a few. Generative AI can also have potentially transformative implications for education research, as in other fields. To realise this potential, policymakers will need to mitigate and manage associated risks - such as those related to access, data privacy, ethics and bias - through sound policy frameworks and effective governance. # Editorial The OECD supports policymakers in making effective and responsible use of generative AI in education. This includes promoting approaches that place human judgement, feedback and oversight at the centre of AI use; strengthening teachers' capacity to engage with AI confidently and effectively; and providing clear, practical guidance on the appropriate use of generative AI in education. The OECD can also foster international co-operation and the exchange of good practices, enabling peer learning across jurisdictions, so that generative AI delivers on its full potential for better learning and more effective education systems. Mathias Cormann Secretary-General, OECD # Acknowledgments This publication is an output of the project on "Smart Data and Digital Technology in Education: AI, learning analytics and beyond" of the OECD Centre for Educational Research and Innovation (CERI) within the OECD Directorate for Education and Skills (EDU). Led by Stéphan Vincent-Lancrin (Deputy Head of Division and Senior Education Economist, OECD), the project team is (or was) comprised of Quentin Vidal (Analyst, OECD), who made key contributions to the project; Yixi Wang (former Secondee, now CNAES, China), who worked on an extensive literature review that informed the report and team's knowledge; Jennifer O'Brien (Assistant, OECD), who managed the final publication process, and; Federico Bolognesi (former Assistant, OECD) who provided excellent project assistance before moving to new responsibilities. The book was edited by Vincent-Lancrin. The overview (chapter 1) was authored by Stéphan Vincent-Lancrin and Quentin Vidal. The four interviews (chapters 5, 6, 10 and 12) were conducted and transcribed by Stéphan Vincent-Lancrin and Quentin Vidal. TThe other chapters were authored by Dragan Gašević and Lixiang Yan (chapter 2); Yuheng Li and Xiangen Hu (chapter 3); Sebastian Strauß and Nikol Rummel (chapter 4); Mutlu Cukurova (chapter 7); Paraskevi Topali, Alejandro Ortega-Arranz and Inge Molenaar (chapter 8); Ryan Baker, Xiner Liu, Mamta Shah, Maciej Pankiewicz, Yoon Jeon Kim, Yunseo Lee and Chelsea Porter (chapter 9); Zachary Pardos and Conrad Borchers (chapter 11); Dominique Guellesc and Stéphan Vincent-Lancrin (chapter 13). Many thanks for very thoughtful contributions and providing feedback on others' chapters! Many thanks as well to Seiji Isotani (chapter 5), Ronald Beghetto (chapter 6), Dorottya Demszky (chapter 10), and Alina von Davier (chapter 12) for their very informative interviews. Within the OECD Secretariat, Andreas Schleicher, Director for Education and Skills and Advisor to the OECD Secretary-General on Education Policy, provided invaluable feedback and is warmly thanked for his continuous encouragement throughout the process. Edmund Misson (Head of the Innovation and Measuring Progress Division, including CERI) is also warmly acknowledged for his continuous support and comments on the draft report. Cassie Hague (Analyst), Hyerim Kim (Analyst) and Anjelica Giordano (Associate researcher) are warmly thanked for their reviews of the chapters, feedback, suggestions and constant willingness to help throughout the production of the report. The EDU communications team led by Joanne Caddy is gratefully acknowledged for their support and delivery under severe time pressure: Duncan Crawford copy-edited the Executive summary and Overview of the book; Della Shin and Sophie Limoges did the layout; Eda Cabbar managed the publication process with the OECD central publications team. Colleagues in the IMEP division and the Director's office are also gratefully acknowledged for their comments, suggestions or friendly encouragement. The CERI governing board is thanked for very helpful comments and feedback on the initial and interim ideas for the book, which they collectively chose, and for their feedback on the final manuscript. Special thanks to the country coordinators of the project for their feedback but also engagement, enthusiasm and sharing of information during the project meetings. Korea and England are also thankfully acknowledged for their financial support to the project. The Gerhard Richter Atelier are warmly thanked for granting permission to use Richter's "Lesende" as the cover image of this book, representing the uncertainty induced by generative AI with a comforting tribute to valued human skills such as reading. The book is dedicated to the late Tia Loukkola, former Head of CERI (and IMEP), a lovely colleague and friend who brightened discussions on the possible impacts of generative AI with her mischievous smile. # Table of contents # EDITORIAL 3 # ACKNOWLEDGMENTS 5 # EXECUTIVE SUMMARY 11 # CHAPTER 1. EXPLORING EFFECTIVE USES OF GENERATIVE ARTIFICIAL INTELLIGENCE IN EDUCATION: AN OVERVIEW 13 What is the general uptake of GenAI? 14 When does GenAI improve learning outcomes? 20 What do educational GenAI tools look like? 26 How could GenAI enhance the effectiveness of education systems and institutions? 28 Concluding remarks 30 Annex 1.A. Examples of country strategies and frameworks on generative AI in education 37 # CHAPTER 2. GENERATIVE AI FOR HUMAN SKILL DEVELOPMENT AND ASSESSMENT: IMPLICATIONS FOR EXISTING PRACTICES AND NEW HORIZONS 39 Introduction 39 Existing practices 41 Challenging assumptions and envisioning new horizons 50 Conclusion 55 # PART 1 ENHANCING STUDENT LEARNING WITH GENERATIVE AI 65 # CHAPTER 3. LEARNING WITH DIALOGUE-BASED AI TUTORS: IMPLEMENTING THE SOCRATIC METHOD WITH GENERATIVE AI 66 Introduction 66 Generative AI meets traditional, AI-powered pedagogical agents 67 Enhanced agent roles and capabilities 68 Pedagogical design and interaction frameworks 70 Working in practice: the SPL demonstration system 73 Framework for efficacy study 77 Challenges, ethics and practical implications 78 Future directions and research roadmap 80 Conclusion 83 Annex 3.A. Technical aspects of educational GenAI agents 87 # CHAPTER 4. FOSTERING COLLABORATIVE LEARNING AND PROMOTING COLLABORATION SKILLS: WHAT GENERATIVE AI COULD CONTRIBUTE 91 Introduction 91 Collaborative learning: Collaborating to learn and learning to collaborate 92 Supporting collaborative learning with generative AI 96 Outlook: Impulses for the future of GenAI in CSCL 104 Acknowledgements 108 # CHAPTER 5. DEVELOPING CREATIVITY WITH GENERATIVE AI: A CONVERSATION WITH RONALD BEGhetto 117 What creativity entails 117 Fast versus Slow AI uses 118 Principles and tools to foster creativity with generative AI 119 Beyond text generation: multimodality and general artificial intelligence 121 # CHAPTER 6. AI IN EDUCATION UNPLUGGED: A CONVERSATION WITH SEiji ISOTANI 122 AI Unplugged 122 AI Unplugged in action 123 GenAI Unplugged 125 # PART 2 AUGMENTING TEACHERS' PERFORMANCE WITH GENERATIVE AI 129 # CHAPTER 7. A CONCEPTUAL FRAMEWORK FOR TEACHER-AI TEAMING IN EDUCATION: HARNESSING GENERATIVE AI TO ENHANCE TEACHER AGENCY 130 Introduction 130 How do teachers use GenAI in education: early benefits and concerns 131 A working definition of teacher agency 132 Three conceptualisations of AI in education and implications on teacher agency 133 Generative AI and teacher-AI teaming 142 Acknowledgements 144 # CHAPTER 8. TRANSITIONING FROM GENERAL-PURPOSE TO EDUCATIONAL-ORIENTED GENERATIVE AI: MAINTAINING TEACHER AUTONOMY 147 Introduction 147 Related work 149 Autonomy vs. automation in AI-driven educational tools 150 The development of a GenAI prototype using design-based research and participatory approaches 152 Discussion 160 Conclusions 162 Acknowledgements 162 Annex 8.A. Description of participant demographics in the study 166 # CHAPTER 9. GENERATIVE AI AS A TEACHING ASSISTANT 167 Introduction 167 Teaching assistants 168 Case study analysis: The JeepyTA platform in universities 171 Other case studies of AI-augmented TAs 183 Reflection on generative AI as teaching assistants: implications and policy recommendations 184 Conclusion 187 # CHAPTER 10. GENERATIVE AI TOOLS TO SUPPORT TEACHERS: A CONVERSATION WITH DOROTTYA DEMSZKY 192 Lesson planning and curriculum material development 192 Classroom analytics 194 Real time support 194 Feedback on student work 197 Real-life implementation 197 # PART 3 IMPROVING SYSTEM AND INSTITUTIONAL MANAGEMENT 199 # CHAPTER 11. AI IN INSTITUTIONAL WORKFLOWS: LEARNING FROM HIGHER EDUCATION TO UNLOCK NEW AFFORDANCES FOR EDUCATION SYSTEMS AND INSTITUTIONS 200 Introduction 200 Emerging opportunities 201 Rationale for adoption and future direction 211 # CHAPTER 12. GENERATIVE AI FOR STANDARDISED ASSESSMENTS: A CONVERSATION WITH ALINA VON DAVIER 215 Enhancing the productivity of item design 215 Improving the assessment of writing and speaking skills 216 High-stakes assessment and next steps 218 # CHAPTER 13. GENERATIVE AI AND THE TRANSFORMATION OF SCIENTIFIC RESEARCH 220 Introduction 220 The use of GenAI in scientific research 221 Effects and challenges of GenAI in scientific research 234 Conclusion 239 # FIGURES Figure 1.1.Increase of ChatGPT users as a share of Internet users,2024-2025 15 Figure 1.2.How do European students use AI to study? (2024) 17 Figure 1.3. Germany: Purpose and frequency of higher education student use of AI for their studies (2025) 18 Figure 1.4. Teachers' use of and opinions about AI in teaching (2024) 19 Figure 1.5. Successfully performing a task with GenAI does not automatically lead to learning 21 Figure 1.6. Educational GenAI tutoring can outperform in-class learning 22 Figure 1.7. Using GenAI can enhance human creativity and writing quality 24 Figure 2.1. VizChat – an LLM-based chatbot designed to enhance the ability of leaders and educators to interpret and understand visual learning analytics. 47 Figure 2.2. Data comics - using LLMs to generate visual feedback based on multimodal data about learning process 47 Figure 2.3. Formative process assessment feedback on self-regulated learning in the FLoRA platform 49 Figure 2.4. Balancing learning gains and performance with GenAI 51 Figure 2.5. User interface of the FLoRA platform for formative process assessment of skills for history taking in medical education 53 Figure 3.1. An overview of the SPL functionalities 74 Figure 3.2. An example of Socratic tutoring session in SPL 75 Figure 3.3. A conversation snapshot demonstrating the adaptability of SPL beyond simply questioning Figure 3.4. The architecture overview of the SPL system 87 Figure 4.1. CSCL design dimensions 94 Figure 4.2. Roles GenAI can assume to support collaborative learning 97 Figure 4.3. Targets of GenAI support and examples 99 Figure 6.1.AI Solution to Support Essay Correction 124 Figure 7.1. The replacement paradigm on the coordinates of teacher agency vs automation 134 Figure 7.2. The complementarity paradigm on the coordinates of teacher agency vs automation 135 Figure 7.3. Transactional teacher-AI teaming 136 Figure 7.4. Situational teacher-AI teaming 137 Figure 7.5. Operational teacher-AI teaming 137 Figure 7.6. Praxical teacher-AI teaming 138 Figure 7.7.Synergistic teacher-AI teaming 139 Figure 7.8. The augmentation paradigm on the coordinates of teacher agency vs automation 139 Figure 8.1. Human-AI automation model 151 Figure 8.2. The design-based research process 152 Figure 8.3. Overview of the envisioned educational GenAI system 156 Figure 8.4. Screenshots of the low-fidelity prototype for Scenario 1 158 Figure 8.5. Screenshots of the low-fidelity prototype for Scenario 2 158 Figure 8.6. Screenshots of the low-fidelity prototype for Scenario 3 158 Figure 8.7. Matrix on teacher autonomy levels as co-participants during tool design and course enactment. 161 Figure 8.8. Teachers' gender, years of experience and courses expertise as collected during phase 1 and phase 2 166 Figure 9.1.JeepyTA guiding students in making up for part of the assignment 173 Figure 9.2. JeepyTA providing feedback on the first step in a student essay assignment – the essay prospectus. 174 Figure 9.3.JeepyTA explaining a key detail about an algorithm 175 Figure 9.4. JeepyTA explaining the errors the student encountered while solving a programming problem and giving advice for diagnosis 177 Figure 9.5.JeepyTA summarising the weekly discussion within the forum 178 Figure 9.6. JeepyTA supporting brainstorming and idea generation in a "games and learning" course 179 Figure 9.7. JeepyTA acting as a Mexican American persona, "Felipe", bringing a specific persona to recommendation 181 Figure 9.8. JeepyTA suggesting discussion questions to start off a weekly discussion on coding qualitative data in the "Quantitative Ethnography and Epistemic Network Analysis" course 182 Figure 10.1. Tutor Copilot: a way to mobilise less qualified tutors effectively, 2024 195 Figure 11.1. Projections of courses at a large public university 202 Figure 11.2. Embedding-based models for mapping problems, skills, and curricula 204 Figure 11.3. Average semester-level credit hours (left) and predicted semester workload (right) for STEM and non-STEM at a large public university in the United States 209 Figure 12.1. A process for human raters to review assessment items generated with GenAI 216 Figure 13.1. Change in AI engagement across all scientific fields 221 Figure 13.2. Proportion of "LLM-modified papers" by discipline 222 Figure 13.3. Uses of AI by researchers, 2025 223 Figure 13.4. A possible interdisciplinary Human-AI collaborative educational research model 229 Figure 13.5. Annual number of scientific publications 237 # TABLES Table 1.1. Examples of different categories of AI 14 Table 8.1. Characteristics of studies applying Human-Centred Design principles in the design of GenAI-based solutions 150 Table 8.2. Selected excerpts of evidence related to participants' use of GenAI. 154 Table 8.3. Selected excerpts of evidence related to participants' ideas about GenAI pitfalls 155 Table 8.4. Prototype description under three use scenarios 157 Table 8.5. Selected excerpts of evidence related to participants' ideas about GenAI pitfalls. 159 Table 13.1. Phases of the research process and capacities of available AI tools 233 Table 13.2. The impacts of GenAI on science: A synthesis table 240 # Executive Summary Generative artificial intelligence (GenAI) is rapidly entering education systems worldwide, raising expectations of more personalised learning, enhanced teaching practices, and more efficient system management. The OECD Digital Education Outlook 2026 draws on the best available empirical research, design experiments, and expert insights to explore where GenAI shows promise, and how education stakeholders can steer its effective and responsible adoption. Evidence shows that GenAI can scale personalised learning support, enhance feedback quality, and automate parts of assessment. But this convenience can come at a cost. When students depend too heavily on GenAI, metacognitive engagement – the mental processes and effort that turns answers into understanding – drops. This results in a misalignment between task performance and genuine learning (chapters 1 and 2). While some studies show both improved student outputs and learning, others do not, particularly when tools provide direct solutions rather than supporting true learning processes. Effectively integrating GenAI into teaching and learning may require that teachers encourage student agency and emphasise process, such as how students think and learn, rather than student output. Hybrid systems that combine GenAI with explicit pedagogical models, such as structured tutoring strategies or evidence-centred assessment design, show more promise than general-purpose chatbots (chapter 2). # Enhancing student learning with generative AI One of the most striking uses for GenAI is tutoring. Unlike the rigid dialogue trees of traditional AI tutors, GenAI can hold flexible, personalised conversations, adapting explanations and language to individual learners' needs. Some AI tutors use methods like Socratic questioning to develop subject knowledge, critical thinking and reflection. The evidence is still emerging, but prototypes show promise (chapter 3). Beyond one-on-one tutoring, GenAI is supporting collaborative learning. Studies identify four main roles: acting as an information hub, generating personalised materials to support group work, providing feedback to teachers, and acting as a peer contributor in group tasks. While evidence so far is limited, some studies find small-to-medium improvements in subject learning and large ones in critical thinking and teamwork (chapter 4). GenAI may also support creativity. Evidence suggests it is most beneficial when used slowly, to support iterative exploration and reflection as opposed to churning out instant content (chapter 5). In this sense, it can also undermine creativity by reducing original thought. Importantly, GenAI has the potential to support students in places with limited digital infrastructure. A large-scale experiment in rural Brazil showed that even with intermittent connectivity and minimal equipment, AI could provide feedback and guidance. Small language models running offline on mobile devices represent a promising avenue for GenAI to bridge digital divides, despite their technical limitations (chapter 6). # Augmenting teachers' performance with generative AI GenAI promises to drastically change the way teachers work in other ways too, including boosting productivity and the quality of teaching. It can already quickly write summaries, design exercises and even offer real-time tutoring support. But there is a risk that overreliance on GenAI could lead to the loss of skills and teaching expertise. A conceptual framework on how humans and AI can work together offers three paths: replacement, complementarity and augmentation. Replacement of some tasks should be assessed carefully to avoid loss of teacher-student interactions. Complementarity is better, pairing human judgment with machine efficiency. But the most effective approach is augmentation through collaborative engagement. In this model, teachers and AI work in tandem, critiquing and refining each other's outputs. This iterative process offers the greatest potential for improved instructional quality while preserving professional judgement (chapter 7). One of the key issues at the moment is that most tools are designed for general use. Off-the-shelf chatbots rarely align with curricula. That is why some argue for purpose-built educational GenAI systems. These tools can be co-created with teachers and students, giving educators control over how machines behave and how students interact with them (chapter 8). For example, this could enable teachers to set the level of "hallucinations" of the tools and give feedback on their student GenAI interactions. Several GenAI tools are already being used to support teachers, especially in the higher education context. For example, some AI teaching assistants can help teachers, teaching assistants, and students across a wide range of instructional tasks while allowing human oversight. Students rated one such tool as comparable to human teacher assistants in clarity, accuracy and professionalism, though weaker in motivation and developmental guidance (chapter 9). Other early evidence suggests that educational GenAI tools can improve online tutoring quality, especially for less experienced teachers. Research also highlights the benefits of AI-generated teaching materials and analytics for effective classroom dialogue. Yet motivation, relationships, and social-emotional learning remain inherently human responsibilities (chapter 10). # Improving system and institutional management GenAI is also streamlining system and institutional management, enabling new forms of classification and recommendations. At the institutional level, GenAI is already reshaping administrative tasks. Embedding-based models can map equivalencies between courses and programmes, making tasks like admissions, career guidance and curriculum analytics faster and more accurate. Large-scale pilots demonstrate high predictive accuracy and efficiency gains, although human AI collaboration remains a must (chapter 11). Beyond feedback, high-stakes standardised assessment is another field where GenAI promises changes. It can generate exam items at scale and design more authentic tasks, such as interactive writing and speaking tasks that mimic real-life communication. By teaming up with AI, teachers can achieve significant productivity gains (chapter 12). GenAI's impact on research is also notable. In natural sciences, it accelerates everything from hypothesis generation to experimental design. The technology is already changing how education research is performed and will potentially improve education systems' outcomes (chapter 13). For example, AI-generated synthetic datasets simulating real education datasets could expand research possibilities and feedback into policy and practice. Ultimately, when designed with strong pedagogy and a human-centred approach, GenAI can do far more than help students complete tasks. It has the potential to deepen student learning, improve teaching practice and streamline institutional management and research. But these benefits come with risks. Overreliance risks turning students into passive consumers and teachers into supervisors. To unlock GenAI's full potential, education must move beyond generic chatbots towards purpose-built tools for education. The thoughtful integration of general-purpose GenAI tools will be essential – for realising the full learning benefits of GenAI and developing students' GenAI literacy for their future careers. The challenge for policymakers is to ensure that GenAI is a learning partner and not a learning shortcut. # 1 # Exploring effective uses of generative artificial intelligence in education: An overview This chapter presents an overview of the findings of the OECD Digital Education Outlook 2026. After a presentation of generative AI (GenAI) and of its uptake in society and education, the chapter shows how research and development on GenAI can inform policy and practice in education. It argues that general-purpose GenAI carries risks for learning, and that it must be used with pedagogical purpose or redesigned as specific educational GenAI tools. A number of educational GenAI tools and their functionalities are presented as examples. GenAI can also support educational workflows within education institutions and systems and present new opportunities for educational research. This report examines generative AI (GenAI), a transformative technology that brought artificial intelligence into the public spotlight, including for students and education policymakers, following the launch of OpenAI's ChatGPT in 2022. Unlike earlier educational AI systems, GenAI is available and used by students outside of educational institutions, with or without the blessing of teachers, school leaders and policymakers. This presents both significant opportunities and complex challenges for education. After clarifying what is meant by GenAI, this chapter gives an overview of the uptake of GenAI among OECD populations, including students and teachers. It then provides a summary of the knowledge and information in this OECD Digital Education Outlook 2026: research evidence on the effects of GenAI on student learning, examples of what educational GenAI could look like, and possible uses to improve workflows at the institution and system levels. # Generative AI in education # What is generative artificial intelligence? GenAI is a subset of AI focused on producing new content such as text, pictures, videos, songs, mathematical equations, computer programmes, typically in response to a question or command ("prompt"). These outputs are generated based on large volumes of training data. To do this, GenAI relies on advanced machine-learning techniques, such as neural networks based on transformers (notably Generative Pre-trained Transformer (GPT)), embeddings, tokens, etc. Most people have experienced GenAI via chatbots based on large language models (LLMs) such as OpenAI's ChatGPT, Google's Gemini, Microsoft's Co-pilot, Anthropic's Claude, Mistral's LeChat or Deepseek's Deepseek. In contrast, non-generative AI systems mainly produce predictions, classifications, recommendations, and ratings, for example for movies, books or other products and services. While they may use similar techniques as GenAI, their primary goal is to identify patterns and relationships in vast amounts of data, rather than create new content. Those AI systems are sometimes referred to as "rule-based", "predictive", or "good old-fashioned" AI. Despite often being less visible to end users, these systems are still powerful and have a variety of uses, including in education. They are embedded in assistive technologies, for example for students with special needs, used to adapt learning to personal needs within intelligent tutoring systems, to score assessments or to predict whether students are at risk of dropping out (OECD, 2021[1]). An important distinction should be made between AI tools that are general-purpose and those that are specialised (in our case, mainly educational): general-purpose systems are versatile and designed to serve many purposes, including educational ones, whereas specialised educational tools are designed for educational purposes only (see Table 1.1). Table 1.1. Examples of different categories of AI <table><tr><td></td><td>Non-generative</td><td>Generative</td></tr><tr><td>General-purpose</td><td>Speech-to-text, Text-to-speech Note-taking tools (image-to-text) AI translation software</td><td>Chatbots (e.g. ChatGPT, Deepseek, Gemini) Image, video or sound generators</td></tr><tr><td>Educational</td><td>Intelligent Tutoring Systems (e.g. Assistments, Lalilo, PILA) Early Warning Systems Simulations (AR/VR)</td><td>GenAI tutors (e.g. Gauth, Khanmigo, Question AI, Socratic Playground) AI Teacher Assistants (e.g. JeepyTA, Coteach, CoTutor)</td></tr></table> # What is so special about general-purpose GenAI tools? General-purpose GenAI tools often provide pertinent and contextualised answers to questions, with the ability to clarify and ask follow-up questions. These capabilities were not possible with earlier (non-generative AI) natural language processing. They are trained on massive data sets that exceed what humans could retrieve manually. Moreover, they are flexible and can be applied to many different subjects. Contrary to most educational AI, general-purpose GenAI tools usually offer free versions, enabling students and teachers to use them even if they are not provided by universities or schools, assuming they have an adequate device and connectivity. Even offline, small language models can run, albeit with lower performance (Isotani, 2026[2]). A series of well-known shortcomings are also specific to current GenAI systems and inherent to its technology. Because they are based on probabilistic models, they can "hallucinate", that is, produce a plausible but wrong answer or fabricate details of an output. They do not generate consistent results over time. For example, repeating the same task several times will yield (at least slightly) different answers or productions, which is sometimes a problem. This is due to regular system updates and to their probabilistic nature. As they are trained on available datasets, their answers and other productions tend to reproduce the views and perspectives represented in those datasets, which are overwhelmingly based on English-speaking (and Western) cultures. For example, unless prompted otherwise, they will typically use Western names or examples in their production. In addition, despite appearing intelligent, GenAI tools do not "understand" the input they process or the content they generate. As a result, their outputs typically require human supervision and scrutiny, often more than specialised, non-generative AI systems. While beyond the scope of this report, GenAI also comes with a series of societal challenges. Many observers are concerned by its environmental footprint, though this is still difficult to measure and compare with other digital technology. The dissemination of AI-generated information and data may decrease the quality of future generated content (as they enter their training datasets) and amplify some current limitations of our knowledge. This will make critical thinking and the development of metacognitive and higher-order thinking skills even more important than before. The full impact of how GenAI might transform societies, labour markets and economies is still emerging. # What is the general uptake of GenAI? Most people experience GenAI through chatbots based on large language models (LLMs), such as OpenAI's ChatGPT, Google's Gemini, Microsoft Copilot, Anthropic's Claude, Mistral's LeChat, and Deepseek's Deepseek-R1. As of April 2025, based on website traffic data, chatbots dominated public use of GenAI tools, accounting for $95\%$ of monthly traffic to the top 60 GenAI platforms. ChatGPT alone represented about $78\%$ of the monthly visits, down from $89\%$ in April 2023 (Liu, Huang and Wang, 2025[3]). Image-generating tools accounted for $2.4\%$ of GenAI websites' traffic, video and audio tools for $1.9\%$ , and productivity and business tools for less than $0.5\%$ . While these shares remain small, the use of these systems has grown significantly since 2023, in line with the overall growth of GenAI use. Competition is also mounting across platforms, with newcomers such as Deepseek and Perplexity gaining market share since 2023. Liu, Huang and Wang (2025[3]) show that the use of GenAI tools has both expanded and intensified. For example, between 2024 and 2025, the number of unique users of ChatGPT grew by $42\%$ , visits per user increased by $50\%$ and the average session duration doubled from 7 to 15 minutes – resulting in the doubling of its traffic ( $113\%$ growth). Most of this growth has been driven by users in high-income countries. In 2025, they accounted for $60\%$ of GenAI use (compared to $55\%$ in 2024), against $39\%$ for middle-income countries and less than $1\%$ for low-income countries (see Liu and Wang $(2024_{[4]})$ for 2024 data). This reflects strong uptake in OECD members as well as accession and key partner countries such as Brazil, China and India. However, it also points toward a widening digital divide based on an adoption and use gap. Part of this gap might be due to measurement issues, as users in low-connectivity regions may not be able to access platforms via the Internet and use versions running offline on their device. Figure 1.1 presents the share of Internet users that used ChatGPT in 2025 and 2024 and thus provides an estimate of the uptake of GenAI tools across populations, acknowledging that averages mask higher usage among younger generations. Figure 1.1. Increase of ChatGPT users as a share of Internet users, 2024-2025 The share of Internet users accessing ChatGPT has increased in OECD, accession and key partner countries Note: ChatGPT is not generally accessible in China. While ChatGPT remained by far the largest GenAI chatbot service, local alternative chatbots tend to be more popular in their countries/regions of origin. The figure highlights the growth of the use of GenAI chatbots in almost all countries. Source: Liu, Yan; Huang, Jingyun; Wang, He (2025). Who on Earth Is Using Generative AI? Global Trends and Shifts in 2025 (English). Policy Research Working Paper; Digital; Washington, D.C.: World Bank Group. http://documents.worldbank.org/curated/en/099856110152535288 # Is it common for students to use GenAI? While there is currently no authoritative comparative data on the use of GenAI by students at different levels of education, several domestic and international surveys provide an initial picture of how widely students use these tools and for what educational purposes. In Switzerland, a 2024 statistically representative survey of 8-18 year-old students points to a steep difference in use depending on age. Around $8\%$ of primary students stated they used GenAI tools at least once a week, $30\%$ in lower secondary, about half in general upper secondary education, and $40\%$ in vocational education. Use in the home followed a similar age pattern (roughly $9\%$ , $33\%$ , $54\%$ and respectively) (Oggenfuss and Wolter, 2024[5]). Including uses less frequent than at least once a week, about $70\%$ of Swiss general upper secondary students use GenAI, and other Swiss pupils use it with a similar age/school pattern as intensive users. In Estonia, a national survey of about 16 000 students found that $74\%$ of lower secondary students and $90\%$ of upper secondary students reported using AI tools to support their studies in 2024, with ChatGPT by far the dominant tool (70% of students use it) (Granström and Oppi, 2025[6]). Beyond national case studies, a cross-country European survey of more than 7 000 12-17 year-olds across seven countries (Germany, Greece, Portugal, Romania, Spain, Türkiye, and United Kingdom) saw high use of Generative AI by students. For example, $48\%$ declared having used ChatGPT in 2024, with almost half of them instructed to do so by their teachers (Vodafone Foundation, 2025[7]). The use of GenAI for higher education students seems to align with the age pattern mentioned above, although statistically representative surveys providing information on this are not yet available. Still, a few studies have surveyed a large number of higher education students (and reweighed their answers to make them more representative). In France, a 2023 study of about 4500 students reported that $55\%$ of higher education students used GenAI tools (Compilatio, $2023_{[8]}$ ). In 2025, the share had increased to $82\%$ (Pascal et al., $2025_{[9]}$ ). In Germany, a survey of over 23 000 higher education students found $94\%$ used AI in 2025, including $65\%$ daily or weekly (Husch, Horstmann and Breiter, $2025_{[10]}$ ). A 2024 international survey of 3 000 higher education students in 16 countries also found that $86\%$ used AI in their studies, including $54\%$ daily or weekly (Rong and Chun, $2024_{[11]}$ ). Evidence suggests that student use of generative AI has moved rapidly from marginal to mainstream since 2022. This is illustrated by looking at the trends among US upper-secondary students – the United States being one of the few places where several surveys were conducted over time. Surveys conducted in 2023 already indicated widespread exposure to GenAI, with around $25 - 33\%$ of secondary students reporting having used GenAI for schoolwork (Center for Digital Thriving, Common Sense Media and Hopelab, 2024[12]). In 2024, comparable surveys suggest a marked acceleration, with close to $50\%$ of middle and high school students reporting some use of AI tools, particularly for homework support, idea generation and explanations of difficult concepts (Impact Research, 2024[13]). In 2025, about $68\%$ of teenagers aged 15-17 reported using AI chatbots such as ChatGPT (Pew Research Center, 2025[14]). The above-mentioned increase in GenAI engagement between 2024 and 2025 (Liu, Huang and Wang, 2025[3]) is also likely driven by younger age groups. In 2024, compared to general Internet users, younger age groups and more educated people drove a substantial share of traffic to these tools, signalling early and concentrated use among teenagers and young adults (Liu and Wang, 2024[4]). There is no reason to believe that their contribution to this share decreased. It is possible or even likely that the early experimenters of 2024 might have transitioned to routine users in 2025.[2] In short, students do use GenAI – a small extent in primary education, a moderate share in lower secondary education, but a majority seem to use it regularly in upper secondary and higher education. While student uptake of GenAI varies by country, the overall trends suggest student use is broadly growing across OECD countries. # What do students use GenAI for? Many students are clearly turning to GenAI tools for academic purposes. However, their primary motivations often center on convenience and efficiency rather than deeper learning. When asked why they use GenAI, according to a number of studies, students typically responded they wanted "cognitive support", such as information, explanations and summaries, or "production support", such as idea generation, drafting, and, perhaps more problematically, solution generation. In Estonia, for instance, grade 6-12 students most often reported using GenAI to achieve better scores, make educational tasks easier, and save time. These uses typically do not support student learning. Common uses include answering homework questions and generating ideas. Lower secondary students more often reported fact-checking, while upper secondary students tended to report summarising specific topics and creating visuals for presentations (Granström and Oppi, 2025).[6]). In most of these cases, the primary motivation was efficiency and convenience (rather than deeper learning). Similarly, in the seven-country European survey mentioned earlier, the most common out-of-school, non-instructed learning uses are obtaining information (56%) and getting explanations of terms and concepts (45%). Nearly one-third (31%) report using AI to provide complete solutions to tasks, while fewer (20%) use it for self-regulatory functions such as structuring personalised learning plans or tracking progress (Figure 1.2). These patterns align with findings from in-depth qualitative interviews with Dutch pupils (Topali, Ortega-Arranz and Molenaar, 2026[15]). What do you currently use AI applications for when learning outside of school and not being instructed by your teachers? Figure 1.2. How do European students use AI to study? (2024) Note: Base: All participants; n = 7000; shown without don't know / prefer not to answer. Multiple answers possible Source: Vodafone Foundation (2025), AI in European Schools: A European Report Comparing Seven Countries, https://skillsuploadjr.eu/docs/contents/AI_in_European_schools.pdf In higher education, students seem to mainly use GenAI tools to search for information, as well as for linguistic tasks such as editing, summarising, paraphrasing, and to a lesser extent drafting (Rong and Chun, 2024[11]). Husch, Horstmann and Breiter (2025[10]) provide the most detailed categories of use and show a largely similar picture, with students primarily using it for general search, idea generation and literature research on the "cognitive" side and for summarising and drafting on the "production" side (with about $22\%$ of regular users) (see Figure 1.3). Interestingly, about $33\%$ of students use GenAI as a "learning partner". Taken together, available evidence suggests that a growing number of students use GenAI for general searches, comprehension and drafting, including as a shortcut to complete tasks and homework. The uses do not seem to be very different in higher education and upper secondary education and tend to reflect the study expectations for students at these different levels. # How do teachers use GenAI? The OECD Teaching and Learning International Survey (TALIS) 2024 provides representative, comparative information on how lower secondary education teachers use AI (OECD, 2025[16]). On average, across OECD countries, $36\%$ of lower secondary teachers' report having used AI in their work in the 12 months prior to the 2024 survey, with very large cross-country variations. Around $75\%$ of teachers in Singapore and the United Arab Emirates report using AI compared to fewer than $20\%$ of teachers in France and Japan. Figure 1.3. Germany: Purpose and frequency of higher education student use of AI for their studies (2025) Results from the CHE University Ranking's Student Survey 2025 Note: N=23 288. The survey includes students enrolled in engineering subjects, as well as in psychology, educational science, German studies, and Romance studies. The students were distributed across 171 different higher education institutions (universities, universities of applied sciences, and cooperative education institutions), including six Austrian universities Students in undergraduate programs from the third semester up to and including two semesters beyond the standard period of study were surveyed. See more details on the CHE website. Ranked by descending order of weekly + daily use. Source: Studierendenbefragung für das CHE Hochschulranking. Hüsch, Horstmann and Breiter (2025).[10]. While the survey does not ask whether they used AI (all kinds) or GenAI, the tasks that teachers report suggest that most uses involve GenAI tools. Teachers primarily use AI for preparation and productivity tasks: on average, $68\%$ report they use it to efficiently learn about and summarise topics they teach, and $64\%$ use it to generate lesson plans. Among AI users, on average $25\%$ report using it to review data on student participation or performance and $26\%$ use it to assess or grade student work (Figure 1.4). In addition, $40\%$ of teachers "agree" or "strongly agree" that AI helps them support students individually, on average. Around $50\%$ agree that AI assists in creating or improving lesson plans, though agreement ranges from as low as $18\%$ in France to as high as $91\%$ in Viet Nam. Seven in ten teachers, on average, believe AI could enable students to misrepresent others' work as their own. Around four in ten teachers agree that AI may amplify biases, reinforce student misconceptions, or compromise data privacy and security. As for the teachers who have not used it, they report feeling overwhelmed by the growing expectation to integrate digital tools in education, which they see as a barrier to using AI in their teaching. This varies markedly across systems, from fewer than $20\%$ in Brazil, Chile, Costa Rica, Italy, Morocco, Türkiye and the United Arab Emirates, to over $50\%$ in Croatia, the Flemish Community of Belgium, Japan and Serbia. On average, three in four teachers report that they lack the knowledge or skills to teach using AI. About half of these teachers also believe that AI should not be used in teaching. In terms of school policy, one in ten teachers reported that their school prohibits the use of AI in teaching. Figure 1.4. Teachers' use of and opi