A competent and reflective use of Artificial Intelligence primarily aims to develop the capacities that enable individuals to assume informed responsibility and make well-founded decisions. It is in this sense that the concept of augmented intelligence is applied: AI is mobilised as a co-agent, serving clearly defined pedagogical, educational or organisational purposes, in order to support teaching, learning or organisational processes – without automating them or replacing human action. Human responsibility remains central in all cases. 

Further information can be found in the strategy, in section 2.3. 

Artificial intelligence offers genuine opportunities for the education system, provided that its use is judicious, clearly framed and guided by explicit pedagogical objectives. In the classroom, when integrated into a clearly structured pedagogical approach, AI can help to enrich and differentiate certain learning processes, as well as to promote equal opportunities and inclusion. 

AI can also help to make learning processes more visible, in particular by encouraging pupils to discuss their reasoning, justify their choices and compare human-produced work with AI-assisted outputs. Finally, AI can also provide valuable support for education professionals, notably by assisting with pedagogical preparation and certain organisational or administrative tasks. Further information on this can be found in the strategy, in section 1.1. 

The integration of artificial intelligence into educational contexts involves risks. These risks stem not only from the technology itself, but above all from the conditions under which it is used, the degree of automation involved, and the role assigned to it within learning processes. The first key issue concerns the risk of excessive cognitive offloading.  Early or poorly regulated dependence on AI can lead to reduced cognitive effort, narrower reasoning and weaker critical thinking. Social and psycho-emotional aspects must also be taken into account, as excessive or insufficiently guided use of digital systems can reduce human interaction and undermine the collaborative and relational dimensions of learning. 

From an informational perspective, another major risk lies in overreliance on generated content. Without explicit competences in verification, contextualisation and critical thinking, pupils may be exposed to forms of misinformation that are difficult to detect. The development of critical thinking, source evaluation and an understanding of bias therefore represents a central challenge for education in the age of AI. The use of AI also raises legal, ethical and institutional challenges. Unregulated use of certain tools may result in breaches of the legal framework, particularly with regard to personal data protection, copyright and academic integrity. Taken together, these risks underline the fact that artificial intelligence is not a neutral tool. It shapes how information is produced, selected and interpreted, and can directly affect the quality, reliability, depth and equity of learning processes. Clear guidance, collaborative learning, targeted skills development (in data and AI) and binding frameworks for use are required to manage risks and enable the responsible use of AI. 

Further information on this can be found in the strategy, in section 1.2. 

Generative AI creates content based on statistical relationships between so-called tokens – that is, word or character units – that appear in the training data. In doing so, the system selects, step by step, one of the most likely next tokens to gradually generate text. This results in texts that often appear as though they were human-written. However, this does not mean that the generated content is accurate. Therefore, the results should always be viewed critically. When a language model (e.g., an LLM) outputs incorrect information, this is referred to as a “hallucination”. 

A comprehensive glossary of key terms (such as LLM, token, etc.) can be found here. 

For better orientation, AI and data competencies can be divided into four interrelated dimensions: 

• Technological understanding 

• Competent use 

• Social and reflective perspective 

• Critical and creative co-creation 

These dimensions serve as reference points for progressive development, according to age, educational contexts and learners’ level of digital maturity. They do not describe a fixed target profile, but rather a process of gradual development. Further information on this can be found in the strategy, in Focus 6. 

The KI Kompass provides verified AI tools that are pedagogically sound, technically secure and compliant with data protection requirements. 

The selection of AI tools for the KI Kompass is based on predefined criteria. The key factors are: 

• Educational relevance: The tool must offer clear value for teaching and learning. 

• Compliance with ethical and legal requirements, particularly regarding data protection (GDPR), copyright and the EU AI Act. 

• Technical security: Reliable and secure use in educational settings must be ensured. 

The use of AI in education requires clear reference points. The overview in section 7.1 of the strategy summarises which practices should be encouraged and which should be avoided. 

Three aspects are highlighted here: 

• Maintain explicit human control over all decisions, particularly those related to assessment, guidance, educational support or organisation; 

• Stimulate creativity, analysis and critical thinking by encouraging pupils to enrich, transform and question AI outputs; 

• Strictly protect data and rights, prioritising authorised tools, secure frameworks and open or legally usable content. These recommendations form a shared general framework for education stakeholders, promoting a responsible, critical and creative use of AI. They do not cover all possible rules.

Certain developmental stages require increased vigilance when dealing with digital tools and artificial intelligence. For younger children, the focus of the AI strategy is therefore primarily on developing basic skills in learning environments that are only partially digital mediated. 

Between the ages of 12 to 16, the use of AI can be specifically guided, clearly framed and pedagogically supported. 

As a rule, a sufficient level of digital maturity is expected by the age of 16. However, this does not mean that AI should only be used from the age of 16 onwards. The use of AI by pupils should be selective, targeted and supported. 

Further information on this can be found in the strategy, in section 3.2. 

AI is used effectively in the classroom when it supports learning processes without replacing thinking, effort and independent work. It serves as a co-agent – that is, as a supportive tool – and not as a substitute for learning or teaching. Its use should always be targeted, reflective and responsible. 

The focus is on pedagogical value: AI is only used if it makes a clear contribution to learning, for example through differentiation, structuring or the promotion of creativity. A balanced approach to teaching and learning combines learning without AI (building basic skills), learning about AI (understanding how it works, its opportunities and risks) and learning with AI (reflective use to support learning). 

At the same time, its use requires a critical and transparent approach: results must be checked and questioned, its use must be disclosed, and responsibility must always remain with humans. The aim is to use AI in such a way that it deepens learning and strengthens learner autonomy and critical thinking. 

The KI Kompass website provides various practical materials (such as lesson ideas, references, etc.) and tools. 

There is no clear-cut or simple solution for determining whether students have used AI for their assignments. Even AI systems are not always able to distinguish between text written by a human and text generated by AI. They frequently generate false-positive results (work written by humans that is incorrectly identified as AI-generated) and false-negative results (AI-generated content that goes undetected). 

Furthermore, AI detectors have no legal validity and cannot provide sufficient evidence of academic dishonesty. 

It is important to use pedagogical and manual strategies to inform such decisions, for example by 

• comparing learners’ previous work, 

• analysing writing style, errors or citation practices, 

• requesting drafts, outlines or learning diaries,  

• quizzes or oral presentations are introduced to assess actual understanding, 

• and holding an open discussion with learners; where appropriate, parents or guardians can also be involved. 

The aim is to assess the coherence of the work, clarify expectations and guide learners towards a responsible and transparent use of AI. 

Further information on this can be found in the strategy, in section 5.4. 

Support is available through: 

• IFEN training courses covering AI and data skills, ethical issues and pedagogical applications, 

• KI Kompass – a platform with practical resources, tools and a glossary of key terms, 

• Communities of Practice, where teachers share their experiences and develop their competencies together. 

A comprehensive collection of AI-related terms with explanations can be found in the glossary. 

To promote the responsible and safe use of AI outside of school, it is important to: 

• Discuss rules and risks: Explain that AI is a tool and not a substitute for thinking. 

• Encourage critical thinking: Check AI results (e.g., by comparing sources) and do not blindly trust the technology. 

• Protect privacy: Never input personal or sensitive data into AI tools. 

• Promote transparency: Label AI-generated content.