C. Tosto1, A. Augello2, G. Città1, M. Gentile1, and S. Perna1
1 Institute for Educational Technology, National Research Council, Palermo, 90146 Italy.
2 Institute for High Performance Computing and Networking, National Research Council, Palermo, 90146 Italy.
Students in the same classroom often need different levels of support. Intelligent Tutoring Systems use artificial intelligence to respond to these differences by adapting activities to each learner. This paper explores teachers’ first impressions of ATLAS, the Adaptive Tutor for Learning and Assessment, and what those impressions suggest about bringing personalised digital learning into schools.
ATLAS selects exercises according to students’ responses and estimated skill levels, updating its understanding of their progress as they work. Its activities guide learners through the process of solving a problem, with optional hints and smaller questions that break a difficult task into manageable steps. The platform is designed to support skills such as critical thinking: for example, judging whether an argument’s conclusion follows from its supporting reasons. Reusable templates allow teachers to generate different versions of exercises while keeping a consistent structure.
Teachers remain responsible for shaping the learning experience. They design activities around educational goals, decide how much guidance students should receive, and use visual progress reports to identify difficulties and plan support. This matters because personalisation can mean that students in one class work at different levels and speeds, requiring teachers to coordinate several learning pathways.
To evaluate the prototype, the researchers collected feedback from 38 Italian teachers after a two-hour training session combining an introduction with practical exploration. Questionnaires asked how easy the platform was to use, how useful it seemed for teaching, and whether participants intended to use it. Open-ended questions invited comments on strengths, difficulties and possible improvements.
The authors report an overall positive assessment across these three areas. Teachers’ comments highlighted the value of adapting exercises to individual needs, offering gradual guidance and encouraging reasoning. Automatic exercise generation was also appreciated, suggesting potential to reduce preparation work. These findings indicate that teachers saw promise in both the learning support and the practical teaching features of ATLAS.
Some participants nevertheless reported technical problems, unclear questions or a need for more time to understand the platform. Suggestions included clearer wording, better readability, a more user-friendly interface, the option to add images and improved accessibility for students with special educational needs. These comments point to concrete ways to improve the system and underline the value of teacher training.
The findings describe initial perceptions after a short training experience. They do not establish sustained classroom adoption or improvements in students’ learning. Only 16–17 teachers answered each open-ended question, and their comments were generally brief, so the qualitative findings also remain preliminary.
The paper’s broader message is that successful adoption depends on how well an intelligent tutor fits everyday teaching. Teachers need reliable, accessible tools, understandable explanations of how activities are adapted, and support for guiding different learning pathways. Designing systems around these needs can help personalised learning become a practical part of school life while keeping teachers’ professional judgement central.
