New Innovations in National Education
Isomiddin qizi Bumayram, Adhamjon o‘g‘li Nasimjon
Artificial intelligence (AI) has become a policy-level priority for education systems worldwide, yet countries differ markedly in how they bring AI tools into schools. This article aims to compare the international regulatory foundations for introducing AI into education with the practical experience of four countries — Estonia, South Korea, Australia and the United Kingdom — and to derive lessons for the education system of Uzbekistan. The study is based on document analysis and comparative analysis of two groups of sources: documents of international organisations (UNESCO guidance for policy-makers, UNESCO guidance on generative AI in education and research, the UNESCO AI competency framework for teachers, and the European Union Artificial Intelligence Act) and national policy documents and official reports of the selected countries. The analysis shows that international documents converge on four common principles: final human responsibility, transparency, data privacy and equal opportunity. National practice reveals three distinct implementation approaches: starting with teacher preparation (Estonia), top-down introduction of a ready-made product (South Korea), and defining principles while leaving decisions to schools (Australia and the United Kingdom). The contrast between Estonia and South Korea is particularly instructive: both countries have advanced digital infrastructure, so the difference in outcomes cannot be explained by technical capacity. Estonia began with teacher training and is expanding its programme step by step, whereas the Korean AI digital textbooks were reclassified as optional supplementary materials within months of their introduction. The findings indicate that the success of AI adoption in education depends primarily on teacher readiness and the gradual, evidence-based character of implementation rather than on technology itself. Based on these results, the article formulates five lessons for Uzbekistan: begin with teacher preparation, pilot before scaling, define principles and competencies rather than mandating specific products, build trust among teachers and parents, and evaluate the quality of generative AI output in the Uzbek language. Particular attention is given to the role of pedagogical higher education institutions in preparing future teachers to use AI tools for pedagogical purposes.
Keywords: artificial intelligence; education policy; international experience; UNESCO; teacher preparation; comparative analysis
The introduction of artificial intelligence (AI) into education has become a policy-level task in many countries. In Uzbekistan, measures to create conditions for the accelerated introduction of AI technologies were adopted in 2021 (President of the Republic of Uzbekistan, 2021), and in 2024 the Strategy for the development of AI technologies until 2030 was approved (President of the Republic of Uzbekistan, 2024). In November 2025, the “5 million AI leaders” project was announced, aiming to train 4.75 million students, 150,000 teachers and 100,000 public servants in AI by 2030 (Kun.uz, 2025). The strategic goals have thus been set; the question now is how to implement them within the education system.
Other countries are following different paths, and their experience — including failures as well as successes — offers valuable lessons. The aim of this article is to compare the international foundations for introducing AI into education and the experience of selected countries, and to formulate lessons for Uzbekistan. The study addresses three research questions: (1) what common principles emerge from international regulatory documents; (2) what implementation approaches can be identified in national practice; and (3) what lessons follow for Uzbekistan’s education system.
Two groups of sources were selected for the analysis. The first group comprises documents of international organisations: UNESCO guidance for policy-makers (Miao et al., 2021), UNESCO guidance on generative AI in education and research (Miao & Holmes, 2023), the AI competency framework for teachers (UNESCO, 2024) and the European Union Artificial Intelligence Act (European Union, 2024). The second group comprises national- level documents and official information about them: Estonia (Eurydice, 2025; Ministry of Education and Research of Estonia, 2025), South Korea (Choi, 2025), Australia (Department of Education, 2023) and the United Kingdom (Department for Education, 2023).
The countries were chosen to illustrate the diversity of implementation approaches. Among the scholarly literature, the systematic review by Zawacki-Richter et al. (2019) is of particular importance: it found that research on AI in education gives insufficient attention to the role of the educator and largely lacks critical reflection on pedagogical issues. Celik et al. (2022), in a systematic review of research on AI for teachers, similarly concluded that the benefits of AI depend on teachers’ competence and that teachers’ perspectives remain under- represented in AI development. Holmes et al. (2019) argue that AI in education should be considered not only as a technology for learning, but also as a subject of learning and a factor that changes what students need to learn. Kasneci et al. (2023) emphasise that the effective use of large language models in education requires competencies and literacies from both teachers and learners, along with clear strategies and pedagogical approaches. Selwyn (2019) warns against the expectation that AI can replace teachers and stresses the social and relational nature of teaching.
The study employed document analysis and comparative analysis. Documents published between 2019 and 2026 were reviewed. Each national case was examined according to three criteria: the starting point of implementation (teachers, students or tools), the pace and scale of rollout, and the presence of guiding principles. The analysis relies on publicly available information and reflects the officially announced content of the programmes.
International regulatory foundations. UNESCO documents view the relationship between AI and education in three dimensions: learning with AI, learning about AI, and preparing learners for life in the age of AI (Miao et al., 2021). The guidance on generative AI recommends a human-centred approach, verification of outputs, data privacy and a minimum age of 13 for independent use (Miao & Holmes, 2023). The competency framework for teachers defines fifteen competencies across five dimensions and three levels of progression (UNESCO, 2024). The EU Artificial Intelligence Act classifies systems used in education for admission, assessment and exam proctoring as high-risk, and prohibits the use of emotion recognition systems in educational institutions (European Union, 2024). Four common principles emerge from these documents: final human responsibility, transparency, data privacy and equal opportunity.
Country experience. In Estonia, the national “AI Leap” (TI-Hüpe) programme was launched in September 2025. The programme begins with teacher training; in its first phase about 20,000 students of grades 10–11 and 3,000 teachers receive free access to AI applications adapted for education. From 2026, the programme is to be extended to new 10th- grade classes and vocational schools (Eurydice, 2025; Ministry of Education and Research of Estonia, 2025).
In South Korea, AI-based digital textbooks for English, mathematics and informatics were introduced in selected grades from March 2025. However, only about one third of schools used them, and in August 2025 the National Assembly removed these tools from official textbook status, reclassifying them as supplementary educational materials to be used at the discretion of schools (Choi, 2025).
In Australia, a national framework for the use of generative AI in schools was adopted in 2023. It is based on six principles: teaching and learning; human and social wellbeing; transparency; fairness; accountability; and privacy, security and safety (Department of Education, 2023). The framework does not prescribe any particular product but sets a common direction for schools.
The UK Department for Education, in its policy paper on generative AI, recognises that the technology can reduce teachers’ workload, while emphasising that responsibility for the accuracy and appropriateness of outputs remains with the professional and the institution using it, and that personal data must be protected (Department for Education, 2023).
Comparison of approaches. The experience of the four countries reveals three distinct approaches (Table 1).
Table 1. Approaches to introducing AI into education
Approach Essence Example Strengths and weaknesses Starting with Teachers are prepared Sustainable, but takes teachers first, then tools are Estonia time provided to students Top-down A ready-made tool is Fast, but fails without product rollout delivered to all schools within a short period
and parents
Common rules are set Australia, Flexible, but Defining instead of a product; United implementation depends principles decisions are made by Kingdom on school capacity schools Source: compiled by the authors.
The difference between the Estonian and Korean experiences is particularly instructive. Both countries have advanced digital infrastructure, so the difference cannot be explained by technical capacity. The difference lies in the order of implementation: in the first case, the process started with teacher preparation and is being expanded step by step; in the second, a ready-made product was delivered to schools within a short period. The revision of the Korean initiative after only a few months shows that the availability of a tool does not guarantee its use.
These findings confirm, at the policy level, the gap identified by Zawacki-Richter et al. (2019): when AI is introduced into education, attention is often focused on technology, while the educator is relegated to second place. The Korean experience illustrates the consequences of this approach, whereas the Estonian programme demonstrates an alternative. This is consistent with the conclusion of Celik et al. (2022) that the benefits of AI are realised through teachers’ competence, and with Selwyn’s (2019) argument that teaching is a social practice that technology cannot simply replace.
Lessons for Uzbekistan. The comparative analysis yields the following lessons:
– implementation should begin with teacher preparation: a tool is effective only when the teacher knows how to use it;
– a gradual approach with piloting is necessary: test on a limited scale first, evaluate the results, then scale up;
– defining principles and competencies is more sustainable than mandating a specific product, because tools change quickly;
– the trust of parents and teachers is decisive: their concerns, above all about children’s screen time and data security, should be discussed in advance;
– the language factor must be taken into account: the quality of generative models’ output in Uzbek should be specifically studied before they are used in education.
Uzbekistan has already taken steps consistent with the first lesson. Within the Day of AI Uzbekistan initiative, based on a programme developed at the Massachusetts Institute of Technology, 1,400 computer science teachers from all regions were trained as AI mentors and, through a cascade model, reached 12,572 educators in 145 districts (New Uzbekistan University, 2026). Scaling such teacher-centred initiatives, rather than rapidly distributing ready-made products, appears to be the most promising path.
A concrete task follows from these lessons for pedagogical higher education institutions: preparing future teachers to use AI tools for pedagogical purposes should become an integral part of curricula. The UNESCO competency framework (UNESCO, 2024) can serve as a basis for defining the content of such preparation.
Limitations. Estonia’s programme has only recently started, and it is too early to draw conclusions about its outcomes; this article analyses its officially announced design, not its effects. Information on the Korean experience relies on media reports. The education systems, financial capacity and cultural conditions of the countries differ from those of Uzbekistan; therefore, their experience cannot be transferred directly, and only general lessons can be drawn. As the field is changing rapidly, the data presented reflect the situation at the time of writing.
An advantage for Uzbekistan is that strategic documents have already been adopted (President of the Republic of Uzbekistan, 2024) and the early experience of other countries is available: there is an opportunity to avoid repeating their mistakes.
The international foundations for introducing AI into education rest on four principles: final human responsibility, transparency, data privacy and equal opportunity. National experience reveals three approaches: starting with teacher preparation, top-down rollout of a ready-made product, and defining principles. The comparative analysis shows that the success of implementation depends not on technology, but primarily on teacher readiness and the gradual nature of the process. The main lesson for Uzbekistan is that introducing AI into education should begin with the targeted preparation of teaching staff, including future teachers. Further research should empirically assess the AI competencies of pre-service teachers in Uzbekistan and the quality of generative AI output in the Uzbek language.