Germany has adopted new guidelines for the use of artificial intelligence in education. Christine Streichert-Clivot, who leads the national body that coordinates education policy across Germany (the KMK), sees this as an opportunity for modern education, but also emphasizes the need for ethical standards.
On October 10, the German Standing Conference of the Ministers of Education and Cultural Affairs (KMK) approved recommendations for dealing with artificial intelligence. The KMK plays a central role in coordinating education policy in Germany, as it coordinates the educational competencies of the 16 federal states in the federal system. The KMK also develops common educational standards and coordinates nationwide measures in vocational education and higher education policy.
As part of the "AI-Humanitas: Enabling AI-Skills" series, Dr. Wolfgang König interviewed Christine Streichert-Clivot about AI in the German education system. Streichert-Clivot is the current president of the KMK and Minister for Education and Culture of Saarland, Germany.
The Decoder: Ms. Streichert-Clivot, the topic of AI in the form of ChatGPT has surprised us all. What are the three most important key learnings from the KMK's perspective?
Streichert-Clivot: Artificial intelligence itself is not really new. It has been a topic in science and research for decades. What is new is its wide availability and everyday use for us as users. The high speed at which AI is developing is challenging, whether it's offerings like ChatGPT or apps that generate images and sounds.
The KMK reacted very quickly to the potential impact of artificial intelligence on educational processes, especially regarding ChatGPT. Since early 2023, initial recommendations for dealing with AI have been published in the federal states, expert discussions have been held, and training opportunities have been expanded. At our autumn meeting in October, we adopted a framework for the use of artificial intelligence in schools.
In some European and non-European countries, bans on use in schools were also discussed in response to the release of ChatGPT. In Germany, however, the federal states have given their schools the opportunity to engage constructively and critically with AI in education from the beginning.
From the perspective of the education ministers, it is particularly important that we focus on school and teaching development towards modern education when integrating artificial intelligence into educational processes. To ensure that teachers and students can access digital learning content, platforms, and other resources in the context of AI, we need to build a nationwide uniform educational infrastructure. This should be reliable, high-quality, and compliant with data protection regulations. At the same time, we should also consider at the European level whether a "Large Language Model" (LLM) specifically tailored to the education sector can be developed. Here in Germany, we want to examine whether such an LLM for the education sector is feasible in a cross-state project called "AIS - Adaptive Intelligent System" over the next two years.
Furthermore, science and practice must work closely together to research and test scientifically sound, age-appropriate, and above all practical approaches to AI applications in schools. This is a basic requirement for the further development of our schools.
Finally, common quality standards, clear data protection guidelines, and ethical principles will help us establish national and European standards for dealing with AI. Schools also need freedom to experiment with AI tools as part of pilot projects. Successful practical examples that emerge in exchange with other schools can serve as a template for future structures, so that AI is integrated into our educational everyday life in the long term and sustainably.
The Decoder: AI chatbots are fundamentally changing learning, it's a bit like allowing programmable calculators in class. Mental arithmetic became less important, but the ability to operate the calculator and estimate its results became central. Do you also believe that AI chatbot skills have become one of the central future skills of our time, as AI will have long-term access to the school knowledge canon and it will then be important for students to have the competence to "tease out" this knowledge from the AI chatbot in an action-oriented way (prompt engineering)?
Streichert-Clivot: Competent handling of AI requires specific skills - digital maturity is crucial for us. We are preparing young people for a life and work world that will be increasingly confronted with AI that will have human competencies in the future.
The Standing Scientific Commission of the KMK (SWK) published a stimulus paper "Large Language Models and their Potential in the Education System" at the beginning of the year, providing scientific guidance on dealing with Large Language Models (LLM) in school education processes. Implementation is up to the states.
From today's perspective, it will be crucial to deal with the reflective and critical use of digital media and the use and effects of AI. Unequal access and lack of action skills would exacerbate social division and educational inequality in Germany. In Saarland, we have introduced the subject of computer science for this reason, among others, and are focusing on comprehensive equipment for all students with tablets from the 3rd grade. We are doing this because the digital world has entered children's rooms and is therefore no longer unthinkable in schools.
In addition to questions of AI use, e.g. a clever handling of prompts, it will therefore be crucial to reflect on AI results, recognize challenges, and be aware of ethical issues. It will continue to need people with their creativity, aesthetic awareness, and emotional empathy. That's what makes us human, and we shouldn't and can't let AI take that away from us.
The Decoder: AI chatbots like ChatGPT cannot be banned and are used by students for a wide variety of tasks, according to the Vodafone Foundation. This ranges from individual information searches to tutoring and exam preparation. How can these advantages already be optimally incorporated into school, training, and studies today?
Streichert-Clivot: Young people, as well as adults, are already using AI-based tools today. It will be about incorporating these even more systematically into schools and educational administration.
Students can be supported in their school and extracurricular learning through AI applications, e.g. as a personal learning tutor, as an intelligent tutorial system, or adaptive learning environment. At the same time, school education always takes place in social space and in interpersonal interaction. Therefore, the use of AI applications must not lead to a loss of importance of learning together.
The use of AI applications in school education underscores the need for a modern task and examination culture. This must be expanded to include examination formats that focus on skills for coping with and shaping the digitized world of life and work. In addition to the so-called future competencies (4C - communication, collaboration, creativity, critical thinking), the new examination formats should also consider AI-related skills to promote successful collaboration between humans and AI.
In Saarland, we have already set this in motion for this school year: Our new performance assessment decree, which we have modernized with regard to the challenges of digitalization, now applies. It now also allows digital tools to be included in performance delivery. This includes, in particular, the performance record "media and material-supported work," which must be written at least once per school year in each written subject. Tools and aids can be, for example, analog or digital reference works, students' own notes, textbooks, digital devices with internet access, word processing programs, drawing software, or AI-based applications.
And we want to go even further here: especially for lesson preparation, AI can provide valuable hints for well-designed lessons oriented towards current content. If AI saves time, teachers have more time for the interpersonally valuable interaction with their students.
The Decoder: What activities are planned by the KMK to ensure that the opportunities of AI chatbot learning are also quickly used in the education system?
Streichert-Clivot: To address the mentioned challenges, the Commission for Education in the Digital World (DigiKom) established the ad hoc working group "Artificial Intelligence" at the beginning of 2023, in which all sixteen states are represented. In its work so far, the ad hoc AI working group has identified five thematic areas with the help of scientific and school-practical expertise and made focus areas: influence and effects of AI on learning and didactics, change in examination culture through AI, professionalization of teachers, need for regulation, and access issues to text-generating AI applications.
Based on these priorities, the states are defining a thematic framework for joint state and state-specific work that should serve as orientation for the use of AI in school education processes, especially for the education administration of the states: The "Recommendation for action for education administration on dealing with artificial intelligence in schools" currently being developed is closely linked to the KMK's supplementary paper "Teaching and Learning in the Digital World" from 2021 and thus to a position already agreed between the states of the KMK strategy on "Education in the Digital World."
The Decoder: How do you think the culture in educational institutions needs to change, since the old roles no longer really apply, as the specialist knowledge can be directly queried from the AI chatbot?
Streichert-Clivot: Successful learning is based on a positive relationship between people, this applies to children and young people among themselves, but also to their relationship with teachers, these are teachers, but also all members of multi-professional teams, e.g. school social workers.
In an ever-changing world where transformation is taking place, educational institutions are more important than ever so that young people can find their way even in uncertain times. In this way, school can be a place of meaningful learning where students can try themselves out in a culture of digitality in a protected and trusting environment. Teachers here are not only those who design appropriate learning arrangements, but especially serve as role models for critical thinking and resilient handling of uncertainty. Even against the background of the daily extent of media use by students, but also as an impulse on the way to digital sovereignty, schools also have a great responsibility as places of lived democracy and sociality. Social, demanding learning can also happen with AI support, but the exchange, the reflection of students among themselves and with their teacher, are core elements of successful learning processes.
Where effectiveness is given, the potential of personalized learning with and through AI should also be integrated into the classroom. School as a place of institutionalized learning can have a quality-assuring effect here and help ensure that comparable educational goals are achieved.
The Decoder: What need for change do you see in teacher training?
Streichert-Clivot: The KMK recommendations "Teaching and Learning in the Digital World" from 2021 already describe an increase in competence development in teacher professionalization to strengthen, among other things, dealing with AI as an example of forward-looking competencies. In addition to practical and technical skills, media pedagogical, media ethical, and media-critical aspects are also taken into account. The adjustments to performance reviews and examination formats described in the area "Change in examination culture through AI" are also incorporated into teacher training.
With the increasing integration of AI in education, it is imperative to further strengthen the aspect of teacher professionalization in dealing with it. In line with the KMK recommendations, corresponding competencies must be systematically and bindingly incorporated into the subject sciences, subject didactics, and educational sciences.
At the Saarland Education Campus, which is responsible for the initial, further, and continuing education of Saarland teachers, corresponding training courses are therefore firmly established. In addition, concepts are currently being developed here to systematically integrate AI into the preparatory service on the one hand, and to structurally address it as a cross-cutting issue in further and continuing education on the other. Because the implications and also the fields of application of AI are actually quite different depending on the subject, school type, or grade level, these concepts are being developed jointly by the actors from initial, further, and continuing education.
The Decoder: With regard to the integration of AI in teaching, there are already many best practice examples, but also didactic models such as the "Gray Box Model of Chatbot Didactics" and an "AI Chatbot Learning Framework." To what extent do you consider such concepts suitable for systematically establishing AI-supported teaching?
Streichert-Clivot: Given the dynamics of technological development, many teachers feel challenged about how and for what they can responsibly integrate AI into teaching. Both models can be a good help here because they sensitize to central questions in AI-supported lesson planning and make it clear that AI is not a surefire success but must be systematically embedded in teaching.
At the same time, both models show that it's not simply about bringing AI into the classroom, but about carefully considering where it offers added value. The rapid and far-reaching development of AI also means that we all need courage to try new things. Such models should only be considered to the extent that they support these processes and do not in turn contribute to uncertainty themselves.
The Decoder: How do you personally use AI chatbots and what do you consider most important?
Streichert-Clivot: I myself use AI chatbots mainly to find information quickly, get answers to specific questions, and get support on various topics. Whether it's about breaking down complex topics into sub-areas or developing ideas for projects: AI chatbots offer me too an efficient way to access knowledge and solve problems.
From my point of view, the most important thing when using AI chatbots is that they are user-friendly and intuitive. The interaction should be uncomplicated and clearly structured so that the desired information can be generated quickly. Transparency is also crucial because I want to know that I'm interacting with an AI-supported system and how my data is being used.
Personalization is another important aspect. A chatbot that takes into account my preferences and previous interactions can give me more relevant and useful answers. Finally, I think it's important that the AI is able to refer to human support if it can't respond satisfactorily to a request.
But what remains important for me: Source criticism remains the main task. Not everything that is produced in seconds is also true and correct. So it can never replace thinking and especially reflecting.
