Ei-iE

Education is facing a number of challenges, such as a shortage of teachers, declining formal student outcomes, and increasing heterogeneity in classrooms. At the same time, the development of artificial intelligence (AI) offers new opportunities for innovation, efficiency, and personalized learning. But, the debate regarding AI in education is often rich and existential. Some view it as a panacea for many educational challenges, others view it as a threat to the very essence of the quality of education. Possibly, a more productive answer considers both viewpoints. Rather than approaching AI with blind optimism or fear, we advocate for a possibilistic view of AI in education. This means acknowledging both the potential and the pitfalls of AI, and recognizing that the educational value of AI does not primarily stem from the technology as such, but from how we use it to support meaningful learning.

Take, for example, one of education’s most persistent challenges: heterogeneity in classrooms. AI tools can cater for this heterogeneity and tailor instruction to individual students, provide targeted feedback, and offer differentiated practice opportunities. Yet, in view of reaching the potential of personalization, clear conditions must be established. Highly recommendable are, among others: agreement upon learning objectives, safe platforms to responsibly implement AI, AI algorithms reflecting established insights on learning and instruction, professional teachers taking appropriate educational decisions considering AI-generated data, and well-prepared self-regulated learners.

AI and the quality of education

Realizing high quality education is important for our future. It empowers the next generation to meaningfully shape our society. Therefore, it is important to reflect on the educational objectives that are at stake. These objectives concern the abilities we want students to develop (including knowledge and skills), while not neglecting the relevance of addressing students’ sensitivity to when and how to apply their knowledge and skills, and their inclination to do so responsibly. In a context where AI is an inseparable part of students' lives, it is more important than ever to empower students with deep domain knowledge and regulatory skills to engage with AI in a critical and constructive way.

From a possibilistic perspective, AI can be assumed to provide opportunities to leverage educational quality if well designed and adequately integrated in educational practice. At the same time, it would be naïve to assume that AI as such will unconditionally improve education. Certain conditions need to be met so that the potential of AI for educational quality can be maximally utilized. Here, we highlight three conditions: supportive policy, professional teachers, and pedagogically sound AI systems.

Supportive educational policy

A coherent vision for AI in education requires clear, guiding policy frameworks that not only outline what AI can do in education, but mainly what is desirable in terms of ethical and sustainable AI integration in education. Supportive for schools are also policies that ensure legal clarity on where and how to use AI in educational contexts, secure and accessible digital infrastructure in schools, professional learning opportunities for school leaders, ICT coordinators, teachers and students, and flexible organizational models that enable schools to not only change traditional educational practices through AI, but also transform the educational system where needed.

Professional teachers

AI can empower teachers, but cannot replace them. Ideally, AI serves as a copilot: augmenting human intelligence rather than replacing it. Used deliberately, AI can function as an exoskeleton and assist teachers before, during and after instruction. Before teaching, it can for example suggest relevant examples and assist teachers in designing the learning environment. During teaching, it can help teachers in deliberately composing groups for team work, flag students who may need extra support, or provide instant feedback on typically occurring misconceptions and learning difficulties. After teaching, teachers can reflect on AI-driven dashboards visualizing student progress, use AI to reflect on follow-up instructional approaches to further support students’ learning, and use AI to assist them in evaluating students’ work. Yet, from our perspective it is crucial that the teacher remains in control. Professional judgment, empathy, and contextual awareness cannot be easily coded into algorithms. For AI to truly add value, teachers need well-developed professional expertise and a healthy skepticism towards AI-systems. AI should support rather than dictate their instructional practice.

Educationally grounded AI-systems

Finally, the AI-tools themselves matter in view of realizing their potential. Schools need theory- and evidence-informed AI-tools designed in line with what we know about how people learn and meeting students’ needs. This means that the development of educational AI applications should not be driven primarily by technological or economic considerations, but should prioritize educational value above all. A way to foster the development of AI-tools for educational purposes is through co-design, where developers collaborate with educators, researchers and policymakers to ensure that what is technologically possible becomes also educationally valuable.

To conclude, if we want to address the challenges schools are facing in a meaningful way, it won’t help to keep education artificially separate from AI developments. We cannot ignore the fact that students – and teachers – are already using AI. To make what is desirable from an educational point of view also practically possible in schools, we need to have a fundamental discussion about what goals we want education to pursue, what this means for the evaluation of those goals, and how we can organize our education in a context of AI. Only then, AI will help to make education even better.