State of the art and practice in AI in education

Analysis

Why This Paper Matters

Recent advances in artificial intelligence have sparked widespread enthusiasm about transforming education, but much of this excitement rests on misconceptions about what AI can actually do and a narrow view of education's purpose. Holmes and Tuomi's review, published in the European Journal of Education, cuts through the hype by systematically examining existing AI in education (AIED) systems and the pedagogic assumptions they embed. With over 950 citations, it has become a key reference for researchers, educators, and policymakers seeking a grounded understanding of AIED.

The paper matters because it challenges the often-unquestioned narrative that AI will automatically improve learning. By developing a typology of AIED systems, the authors reveal that different technologies serve very different educational goals—some behaviorist, some constructivist, some sociocultural. This matters for practitioners who must choose tools aligned with their teaching philosophy, and for developers who may inadvertently impose narrow pedagogical models. Moreover, the authors highlight that many expectations are based on misunderstandings of current technical possibilities, urging a more realistic appraisal of AI's capabilities.

Finally, the paper situates AIED within broader societal functions of education, arguing that exceedingly narrow views—such as reducing education to skill acquisition or economic productivity—can lead to misguided AI implementations. This critical perspective is essential as AI becomes more pervasive in classrooms and learning platforms.

Technical Contributions

• Typology of AIED systems: The authors categorize AI applications in education based on their pedagogic and educational assumptions, providing a structured way to compare and contrast different approaches.
• Analysis of pedagogic grounding: They show how AIED systems are grounded in different interpretations of what AI and education is or could be, linking technical design to educational philosophy.
• Identification of roadblocks: The paper discusses potential obstacles on the AIED highway, including technical misunderstandings, lack of knowledge about state-of-the-art AI, and narrow views of education's role in society.
• Critical review: It synthesizes existing AI systems in education, offering a comprehensive overview of the current landscape and its limitations.

Results

As a review paper, it does not present experimental metrics but instead offers qualitative findings. The key result is the typology itself, which classifies AIED systems and reveals that many current applications are based on behaviorist assumptions, while others attempt more constructivist or sociocultural approaches. The authors also find that expectations for AI in education often exceed technical realities, and that narrow views of education can lead to AIED systems that fail to address broader educational goals. No quantitative comparisons are provided, but the paper's citation count (957) indicates its significant influence in the field.

Significance

The paper's significance lies in its critical and structured examination of AIED, which helps align AI development with educational values and realistic capabilities. It provides a common language for discussing AIED systems and their assumptions, fostering more informed design and implementation. By highlighting roadblocks, it warns against overpromising and underdelivering, which can erode trust in educational technology. Ultimately, it encourages a more holistic view of education that considers societal functions beyond mere efficiency or skill acquisition, shaping future research and policy in AIED.