Educational AI at the dawn of responsible governance

Introduction

We are living in a pivotal moment. The rise of artificial intelligence (AI) is no longer a distant horizon, but a daily reality that is transforming the way we learn, transmit, and cooperate. Current trends invite us to look beyond immediate uses to anticipate technological developments, understand how AI integrates into educational practices, and consider the contours of responsible governance. This reflection is part of a strong conviction: AI must not be a simple technical optimization, but a force serving social ties, the development of talent, and fundamental knowledge .

Major trends in AI in education

Recent innovations herald a profound transformation of the educational landscape. Large language models (LLMs) , capable of generating text, coding or simulating complex dialogues, open the way to unprecedented personalization of learning (Kasneci et al., 2023). Added to this is the rise of multimodal generative AI , capable of processing texts, images and sounds, offering more immersive educational environments.

But behind the innovation lies a societal question: what kind of education do we want to build with these tools? While AI can automate certain cognitive tasks, it must not become a substitute for critical thinking or human creativity (Selwyn, 2019). The risk of increased dependency, sometimes referred to as “technological solutionism,” lies in delegating the control of knowledge to opaque systems. This is a major challenge for institutions: ensuring that AI reinforces fundamental learning and stimulates critical thinking , rather than weakening the intellectual training of students.

Educational foresight therefore invites us to design systems where AI acts as a mediator and amplifier , but never as the sole repository of knowledge. Basic knowledge remains the foundation on which all specialization is based, and AI cannot reduce its importance.

Integrating AI into educational and professional practices

The integration of AI into education should not be seen as a superficial addition, but as a lever for pedagogical transformation . Concretely, several axes are emerging:

1. Personalization of courses : Thanks to data analysis, AI can adjust content and learning pace. This adaptability is a major asset in promoting inclusion, particularly for students with specific needs (Luckin, Holmes, Griffiths & Forcier, 2016).

2. Support for active learning : AI can encourage questioning and collaboration rather than just providing answers. Used as a dialogue partner, it enhances the interactive dynamics of learning.

3. Teacher support : far from replacing the teacher, AI can alleviate certain burdens—formatting correction, preparation of materials, administrative follow-up—to allow the teacher to concentrate on the essentials: the educational relationship and human support.

4. Broadening cross-curricular skills : the rational use of AI requires new skills: critical thinking, digital culture, ethical information management. These are becoming educational objectives as important as disciplinary mastery (OECD, 2021).

   
   

Thus integrated, AI can be a tool for emancipation , helping to promote talents and strengthen social ties within educational communities.

Towards responsible governance of educational AI

Any integration of AI into education raises the question of governance . Current models, often developed by large private players, pose challenges of fairness, transparency, and digital sovereignty. Education cannot be satisfied with a consumerist use of tools; it must define appropriate regulatory frameworks.

Responsible governance is based on several principles:

• Transparency and explainability : Users need to understand how models produce their answers and what data are used (Floridi & Chiriatti, 2020).

• Ethics and inclusion : This involves ensuring that AI serves the diversity of students and does not reproduce existing biases (UNESCO, 2021).

• Collective participation : teachers, researchers, students and institutions

• must be associated with the definition of uses, in a logic of shared governance.

• Ecological sustainability : The environmental footprint of large language models is not negligible. Recent work shows that their training generates significant CO₂ emissions (Patterson et al., 2021). Developing energy optimization and sobriety practices is therefore a priority.

Without this framework, AI risks reinforcing social and educational divides. With it, it can become a force for cohesion and democratization of knowledge .

A forward-looking vision: education as a renewed social pact

If we consider AI solely as a technology, we miss its true scope: that of a civilizational tool . Education, at the heart of this transformation, must remain faithful to its primary mission: to educate free, critical, and supportive citizens . AI, when governed responsibly, can contribute to a new educational social pact : a pact that values cooperation, the recognition of talent at all levels, and the awareness that every educational action contributes to the collective interest.

The prospective scenarios for 10 or 20 years ahead should not frighten us, but should invite us to lay the foundations for responsible, shared, and ethical governance today. Thinking about education with AI means agreeing to combine innovation and humanism : a demanding, but necessary, path.

Conclusion

The future of educational AI is not only being played out in technical laboratories, but in the collective choices we make today . Between fascination and fear, it is up to us to chart a path where innovation serves humanity. In this open letter, we call for a shared responsibility: to make AI a lever for emancipation, not a tool for dependency. Education has always been a profoundly social act; with AI, it can also become a profoundly forward-looking and ethical act .

---

References:

1. Floridi, L., & Chiriatti, M. (2020). GPT-3: Its nature, scope, limits, and consequences. Minds and Machines, 30 (4), 681-694. https://doi.org/10.1007/s11023-020-09548-1

2. Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., … & Kasneci, G. (2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences, 103 , 102274. https://doi.org/10.1016/j.lindif.2023.102274

3. Luckin, R., Holmes, W., Griffiths, M., & Forcier, LB (2016). Intelligence Unleashed: An argument for AI in Education . Pearson.

4. OECD. (2021). AI in education: Challenges and opportunities for sustainable development . OECD Publishing.

5. Patterson, D., Gonzalez, J., Le, Q., Liang, C., Munguia, LM, Rothchild, D., … & Dean, J. (2021). Carbon emissions and large neural network training. arXiv preprint arXiv:2104.10350 .

6. Selwyn, N. (2019). Should robots replace teachers? AI and the future of education . Polity Press.

7. UNESCO. (2021). AI and Education: Guidance for policy-makers . UNESCO Publishing.