{"slug": "margit-sutrop-ai-leap-requires-technological-courage-and-educational-wisdom", "title": "Margit Sutrop: AI Leap requires technological courage and educational wisdom", "summary": "A University of Tartu Centre for Ethics working group published an analysis in mid-September identifying 12 risks in three categories — pedagogical, societal, and infrastructure and governance — tied to Estonia's AI Leap (TI-Hüpe) education program and its AI-based learning application. The analysis concludes the greatest risks stem not from the technology itself but from how it is used and the educational and policy choices made, with responsibility shared among AI Leap, schools and school operators, government agencies, research institutions, and technology providers. The report cites an early 2024 Estonian study in which 90 percent of upper secondary school students had used AI in their studies, mostly free versions of commercial applications, and a fall 2024 study in which 47 percent of teachers said they had not used AI.", "body_md": "# Margit Sutrop: AI Leap requires technological courage and educational wisdom\n\nAs Estonia brings artificial intelligence into its schools through AI Leap, its success will depend not on how widely the technology is used, but on whether it strengthens learning while safeguarding teachers' role, educational values and the Estonian language and culture.\n\nIn mid-September, a working group at the University of Tartu Centere for Ethics published an analysis of the risks and value choices associated with the AI Leap (TI-Hüpe) education innovation program and an AI-based learning application. The analysis describes 12 risks related to the use of artificial intelligence in education and offers recommendations on how AI Leap, the state, schools, research institutions and technology providers can mitigate them.\n\nThe analysis was based on the premise that introducing an AI-based learning application in Estonian schools is not merely a technological innovation. It has implications for learning, teachers' work, educational equality, the Estonian language and culture, as well as our system of values. Ultimately, it is also tied to the question of what kind of people and society our education system shapes.\n\n**Risks not confined to technology**\n\nThe 12 risk scenarios described in the report fall into three categories: pedagogical, societal and infrastructure and governance risks. This breakdown shows that the challenges associated with the use of artificial intelligence are not limited to technical performance and cybersecurity. Among other things, they concern the goals of learning and the reliability of assessment, the role of teachers, the sustainability of the Estonian language and culture, data protection and the reliability and sustainability of technological solutions.\n\nThe purpose of the risk analysis is not to sow fear or put the brakes on the use of artificial intelligence. On the contrary, risks need to be analyzed so that innovation can succeed. The risk scenarios described are possible developments that can be prevented if the various stakeholders recognize their responsibilities and make informed decisions. AI Leap is already working to mitigate some of the risks, while for others, agreement is still needed on who is responsible for implementing which measures. Responsibility is shared among AI Leap, schools and school operators, government agencies, research institutions and technology providers.\n\nIn addition to the risks, the Center for Ethics working group also mapped out the value choices involved to make clear what is at stake in the development and use of an AI-based learning application. One of the analysis's most important conclusions is that the greatest risks do not stem from the technology itself, but from how it is used and from the educational and policy choices that are made.\n\n**Will learning become deeper or merely faster?**\n\nA [study](https://www.frontiersin.org/journals/education/articles/10.3389/feduc.2025.1688092/full) conducted in Estonia in early 2024 showed that 90 percent of upper secondary school students had used AI in their studies, mostly free versions of commercial applications, to complete schoolwork more quickly. When a student uses artificial intelligence to quickly produce the required result, that does not necessarily mean learning has taken place. A student may submit a flawless text without being able to independently reconstruct its reasoning, justify its claims or apply the knowledge in a new situation. In this way, learning can imperceptibly be replaced by the appearance of learning. The AI Leap education program was created precisely to prevent this kind of development. However, AI Leap is not only a program to develop a learning application for students, but also to develop teachers' AI competencies. The need is clear: In a [study](https://www.frontiersin.org/journals/education/articles/10.3389/feduc.2025.1622240/full) conducted in fall 2024, 47 percent of teachers said they had not used AI tools in their work.\n\nThe Socratic learning application being developed for Estonian schools through a collaboration between OpenAI and Estonian researchers differs from a conventional chatbot. Its purpose is to use questions, hints and feedback to help students arrive at solutions themselves, organize their thinking and plan their own learning.\n\nThis is a noble goal. A Socratic learning application could make learning deeper, provide learners with more personalized support and make high-quality learning assistance more accessible. But no technology can, by itself, ensure that these possibilities are realized. At least as important is how the application is integrated into the learning process, what kinds of assignments it is used for, how much insight teachers have into students' learning processes and what support is provided to schools.\n\nThe use of artificial intelligence also requires changes to assessment. In the age of AI, academic integrity cannot be protected through prohibitions and monitoring alone. Assignments need to be designed in ways that make students' thinking processes visible, clear rules need to be established for permitted uses and young people need to be taught to disclose their use of artificial intelligence honestly, verify its responses and take responsibility for them.\n\nThe most fundamental question concerns the goals of education in Estonia. Education must prepare people to manage their personal lives, participate in society and succeed in working life. This requires competencies comprising knowledge, skills and values. An AI-based learning application can help develop some of these competencies. Developing critical thinking, communication and collaboration skills and a readiness to participate responsibly in society, however, remains the responsibility of schools.\n\n**AI must not become a parallel teacher**\n\nA second major set of questions concerns the learning relationship between teachers and students. Artificial intelligence can serve as an always-available, patient and personalized guide. For precisely this reason, students may also come to see it as having an authority that it was never deliberately given.\n\nArtificial intelligence is not a teacher, a friend or a moral authority. It does not know the student as a person, nor does it bear responsibility for the student's development. In addition to imparting knowledge, a teacher's role is to inspire students, set boundaries, recognize sensitive situations and discuss values with students — all within a safe and trusting learning environment.\n\nFor this reason, teachers must retain their pedagogical leadership in the implementation of AI Leap. The learning application must support a teacher-led learning process rather than becoming an invisible, parallel pedagogical authority alongside it.\n\nTeachers cannot be held responsible for the consequences of using a tool if they lack sufficient insight, decision-making authority and the necessary support. The current solution does not yet give teachers sufficient insight into students' learning processes. Teachers are being introduced to the possibilities offered by artificial intelligence and are being supported in experimenting with it and sharing their experiences. But they also need subject-specific instructional support, guidance on adapting assessment practices and time to develop shared rules.\n\nThe responsibilities of schools, school operators, the state, the AI Leap Foundation and technology partners must also be clearly defined. Schools remain responsible for the quality and safety of the learning environment and for protecting students' rights. That responsibility does not disappear simply because the introduction of the application is being coordinated by a foundation or the technological solution is provided by a private company.\n\n**AI is not value-neutral**\n\nOne of the less visible but more profound risks is value drift. The responses generated by large language models are shaped by training data, developers' choices, safety filters, content policies and commercial objectives. The risk does not necessarily involve malicious influence. The impact on values can be gradual and difficult to detect, manifesting itself in the repeated selection of examples, the way problems are framed, patterns of argumentation and implicit assumptions.\n\nThis raises important questions: Whose values does artificial intelligence reflect? Are they compatible with Estonia's constitutional values and the fundamental values of its education system? How does the application respond in situations where there are legitimate conflicts of values within society and no single indisputable answer exists?\n\nAligning artificial intelligence with values does not mean programming a single moral doctrine into a machine and making it binding on everyone. In a democratic society, a plurality of values is inevitable and often necessary. Human rights and human dignity must, however, be protected, harm and discrimination must be prevented and significant value choices must be made transparently and democratically rather than through the default settings of a foreign technology company.\n\nIt is particularly important to prevent students from becoming alienated. Artificial intelligence can give the impression of being an empathetic and supportive conversational partner: It is always available, responds patiently and encourages the user. As a result, students may begin to prefer interacting with artificial intelligence even in situations where they need support — or constructive criticism — from a teacher or fellow students. This could reduce teachers' contact with students and their ability to recognize when students are struggling. Education must preserve space for discussion, collaboration, making mistakes and human connection.\n\n**Role of Estonian needs to be more than a user interface language**\n\nAI Leap is also a language and cultural policy choice. International large language models may not draw sufficiently on Estonia's national curriculum, high-quality Estonian-language learning materials, correct terminology or the local cultural context. As a result, Estonian could be reduced to merely the language of the user interface, while more complex concepts, examples, explanations and forms of argumentation are drawn predominantly from English-language data. In the longer term, this concerns the substantive autonomy of Estonia's education system and the continued viability of Estonian as a language of learning, thought and knowledge creation.\n\nWe cannot expect artificial intelligence to produce high-quality Estonian if its development cannot legally make use of high-quality Estonian-language learning materials, literature and professionally edited media content. Copyright-compliant solutions must therefore be found that allow the best content from Estonia's linguistic and cultural sphere to be used both in developing the learning application and in assessing its quality.\n\nAssessing linguistic quality, factual accuracy, cultural appropriateness and alignment with values must be an ongoing process rather than a one-time exercise. This requires the sustained involvement of Estonian universities, researchers, teachers and subject-matter experts.\n\n**Risk analysis must become a tool of managing AI Leap**\n\nThe success of AI Leap should not be measured solely by how many schools, teachers or students use the application. A large number of users may indicate that the application has been adopted quickly, but it does not tell us whether learning has actually become deeper and more meaningful. Creating a Socratic learning application alone is therefore not enough. Schools must teach students to use artificial intelligence in ways that support both learners and teachers. This also requires redesigning learning tasks and assessment. Alongside earning good grades, the goal must be to develop the competencies needed for life, independent thinking and learners' ability to manage their own learning.\n\nIn assessing the program's impact, we need to ask whether students' independent and critical thinking improves, whether educational disparities decrease or increase, how teachers' workloads and professional autonomy change, whether a relationship of trust between teachers and students is maintained and how the application affects the Estonian language and culture.\n\nA risk analysis is not a document to be put on a shelf after publication. It should become a tool for managing AI Leap. Efforts to mitigate risks and the program's actual impact must be assessed regularly, with changes made as necessary to the application, its rules of use and the way the program as a whole is organized.\n\nEstonia has set an example for other countries with its technological boldness. In education, that boldness must be accompanied by responsibility and a willingness to change how technology is used based on evidence. The true test of AI Leap is not how quickly the new application reaches schools or how widely it is used. We can call it a success when AI helps make learning deeper, supports teachers' professionalism and serves the goals of Estonian education without losing any of what makes education valuable in the first place.\n\nThe risk analysis and its main recommendations are available on the [AI Leap website](https://tihupe.ee/en/ai-leap-risk-assessment/) both in Estonian and English.\n\n--\n\nEditor: Marcus Turovski", "url": "https://wpnews.pro/news/margit-sutrop-ai-leap-requires-technological-courage-and-educational-wisdom", "canonical_source": "https://news.err.ee/1610140738/margit-sutrop-ai-leap-requires-technological-courage-and-educational-wisdom", "published_at": "2026-09-18 07:20:00+00:00", "updated_at": "2026-09-18 07:23:49.205702+00:00", "lang": "en", "topics": ["ai-policy", "ai-ethics", "artificial-intelligence"], "entities": ["University of Tartu Centre for Ethics", "AI Leap", "TI-Hüpe", "Estonia", "Margit Sutrop"], "alternates": {"html": "https://wpnews.pro/news/margit-sutrop-ai-leap-requires-technological-courage-and-educational-wisdom", "markdown": "https://wpnews.pro/news/margit-sutrop-ai-leap-requires-technological-courage-and-educational-wisdom.md", "text": "https://wpnews.pro/news/margit-sutrop-ai-leap-requires-technological-courage-and-educational-wisdom.txt", "jsonld": "https://wpnews.pro/news/margit-sutrop-ai-leap-requires-technological-courage-and-educational-wisdom.jsonld"}}