Artificial Intelligence has already become part of students’ everyday learning, often without guidance or a clear pedagogical framework. Pupils use it to solve exercises, organise their study and search for information, while teachers are expected to manage a technology that is evolving faster than schools’ institutional reflexes. The question is no longer whether AI will be used in education, but under what conditions and with what pedagogical impact.
In this context, the first data emerging from pilot applications of AI-based educational tools in Cyprus are particularly significant. These initiatives aim to shift the discussion from theory to classroom experience. The evaluation of their educational impact was recently presented at a seminar at the Cyprus Pedagogical Institute, marking the completion of the European co-funded project Artificial Intelligence in Education – AI4EDU.
From design to the classroom
The AI4EDU project set out to explore, develop and pilot conversational AI educational assistants designed to support students and teachers in the learning process. These applications were not conceived as generic AI tools, but were developed with a clear pedagogical focus, supporting teaching without replacing the role of the educator.
As part of the project, conversational assistants for both students and teachers were developed for use in the classroom or during independent study. Their purpose was to guide thinking, provide feedback and support understanding. The tools were created through a collaborative design process involving researchers and teachers, ensuring alignment with the realities of classroom practice. Pilot implementations took place in authentic school settings during the final phase of the project, throughout the 2024–2025 school year.
What emerged
Analysis of data collected during the pilot phase made it possible to draw concrete conclusions about how Artificial Intelligence functioned in the classroom, when it enhanced learning and under what conditions its limitations became apparent.
The findings provide a realistic picture of the introduction of AI into educational practice, highlighting both its potential and its constraints as observed during the pilots. In particular, it emerged that:
-
Artificial Intelligence did not function as a provider of “ready-made answers”, but as a tool to support thinking. Conversational assistants guided students through questions and prompts, helping them develop reasoning rather than seek immediate solutions.
-
The use of AI strengthened reflection and active engagement. Pupils were able to review their steps, identify mistakes and reconsider responses before turning to the teacher.
-
AI was most effective when it was organically integrated into the lesson rather than used as a stand-alone digital tool. When linked to specific learning objectives, students used it more meaningfully, while the absence of a clear framework limited its impact.
-
The role of the teacher proved decisive. Educators who clearly defined when, how and why AI would be used, and who intervened when necessary, succeeded in integrating the technology without it replacing teaching.
-
The experience of using AI varied by subject and student level. In subjects involving more complex concepts, AI was mainly used to support explanation and understanding, while elsewhere it functioned more as a tool for organising thinking and study.
-
In some cases, the lack of clear instructions led to difficulties. When students were unsure how they were expected to use the application, its use became fragmented or ineffective, underlining the need for clear frameworks and shared classroom rules.
-
The pilots highlighted the need to strengthen AI literacy. Both teachers and students required support to understand the capabilities and limitations of AI systems in order to use them responsibly.
-
AI did not function uniformly across classrooms. Outcomes varied depending on how it was embedded in teaching, the subject area and the role of the teacher, showing that AI is neither a solution nor a threat in itself, but a tool that requires pedagogical judgement.
The bigger picture
As became clear in the discussion following the presentation at the Cyprus Pedagogical Institute, the pilot findings do not point to a uniform or automatic impact of Artificial Intelligence on learning. Instead, AI’s contribution depends largely on how it is integrated into teaching and on the pedagogical choices that precede its use.
Where there was a clear framework and an active role for the teacher, AI tools functioned supportively, strengthening students’ thinking and participation. Where use was not accompanied by guidance, difficulties and ambiguities emerged around their role in the classroom.
At the same time, concerns were raised about the need for shared rules of use, basic AI literacy and institutional support, so that the integration of Artificial Intelligence is not fragmented but pedagogically grounded.