Three-Hour Workshop, New Perspectives: Teachers Explore AI for Learning

The programme, called INSIGHT, combined basic explanations of AI, practical activities and guided discussion about its educational possibilities and limitations.

A teacher preparing one lesson may need several versions of the same material, each suited to students with different reading skills, language backgrounds and learning needs. Artificial intelligence offers possible help with that workload, raising a practical question: does showing teachers how it works make them more willing to use it?

Researchers Janne Mesenhöller and Katrin Böhme at the University of Potsdam explored that question in 'Empowering teachers to use AI for differentiation: changes in teachers' acceptance of AI-based technologies following participation in a short professional development training session,' published in Frontiers in Education. Their pilot study found that teachers saw AI as more useful and easier to use after a three-hour workshop, with no change in their intention to use it.

The Classroom Challenge Behind the Research

Differentiation means adjusting teaching to the needs of different learners, through changes to task difficulty, explanations, feedback or the time available for learning. It can support students who need extra help and those ready for more demanding work, with adaptations made for individuals, groups or an entire class.

The paper cites a separate German survey in which 83% of teachers said that responding to different learning needs created considerable additional work, and nearly half felt overwhelmed by those demands. AI tools could help teachers produce texts at different reading levels, translate materials, generate questions with varying difficulty and offer personalised feedback. Large language models, the technology behind tools such as ChatGPT, have made these possibilities accessible without the specialised software required by some educational AI systems.

The researchers wanted to understand whether a short, practical training session could improve teachers' acceptance of these technologies within the time constraints of everyday school life.

What Teachers Did During the Three-Hour Workshop

The programme, called INSIGHT, combined basic explanations of AI, practical activities and guided discussion about its educational possibilities and limitations.

Teachers explored core concepts through familiar examples, including the distinction between machine learning and deep learning. Small groups used Google's Teachable Machine to train a model to recognise two hand gestures: pointing and a high five. Collecting images, training the system and testing its responses gave participants a concrete demonstration of how performance depends on the data provided.

The workshop introduced large language models through examples connected to lesson preparation and classroom learning. Teachers examined ways to adapt reading materials, translate content and create tasks at different levels, including a biology question about photosynthesis rewritten for three levels of competence.

Discussion covered student uses, such as requesting repeated explanations or receiving feedback on the clarity and structure of a piece of writing. Teacher guidance remained central, including recognition that students differ in their ability to work effectively with AI and may need support to participate fairly.

INSIGHT ran in person at five schools in Brandenburg, Germany, between March and September 2023. The same two trainers delivered the sessions using consistent materials and activities. The final analysis included 100 teachers with matched questionnaires completed immediately before and after training, covering primary, lower secondary and upper secondary education.

Teachers Saw More Value Without Stronger Plans to Use AI

The questionnaires measured three aspects of technology acceptance: perceived usefulness, perceived ease of use and behavioural intention, meaning teachers' stated intention to use AI.

Responses used a four-point scale, from strong disagreement to strong agreement: Average usefulness scores rose from 2.89 before training to 3.02 afterwards. Ease-of-use scores increased from 2.85 to 2.99. Both changes were significant, with small effect sizes, suggesting a modest improvement in perceptions. Teachers' intention to use AI shifted from 3.07 to 3.01. The findings do not establish either an increase or a decline in willingness to use the technology.

Participants rated the workshop positively: 88% found it personally helpful and 78% considered it useful for future lesson planning and instructional design. Its broader relevance attracted high agreement, with 95% seeing value across subjects, 91% across school types and 85% across grade levels. The gap between general relevance and personal usefulness matters because recognising that AI has educational potential does not necessarily mean a teacher sees a clear place for it in their own lessons.

Teachers already held reasonably favourable views before the workshop, leaving less room for improvement, particularly in their intention to use AI. The authors suggest that limited time, inadequate digital infrastructure and unclear school rules could help explain the unchanged intentions.

What This Means for Schools and Future Training

A three-hour session appears to be a promising way to introduce teachers to AI through examples connected to their work. The evidence supports a careful reading: participation was associated with more positive perceptions, and teachers generally found the experience relevant.

Questionnaires captured immediate reactions, leaving open whether those perceptions lasted or led to changes in classroom practice. The research did not measure actual AI use, improvements in teaching or student learning outcomes. Its questions concerned AI technologies broadly, so the results cannot be treated as a specific measure of acceptance of large language models.

The authors call for larger, more diverse studies with comparison groups and longer follow-up periods, including investigation of school policies, technical resources and institutional support. Observing teachers' actual use would help clarify what turns a useful introduction into a lasting classroom practice.

Elements of INSIGHT have been incorporated into freely accessible learning resources through the KISS-Pro project, including a self-paced course on the openHPI platform.