Generative AI and Classroom Practice – What is the story?

Preamble

I have spent the past few months engaged in a number of research activities associated with the use of AI in schools and in further education and training (FET) contexts in Ireland and across the EU. This work has consisted of research on AI uses in Irish schools with UNESCO and through H2’s involvement in a number of Erasmus+ projects. This work has afforded me the opportunity to see how schools and FET institutions are, or are not, using AI, in classrooms or learning settings with learners. In this blog post I want to share some high-level observations on this work, which are drawn from a range of sources that include existing policy documents, reports and recent research on classroom practice.

We are primarily talking about generative AI (GenAI)

In the Irish context when we typically speak about AI in schools or education, we are talking about generative AI (GenAI). A recent EU policy report, Friend or foe? Evidence from the use of generative artificial intelligence in learning and teaching, defines GenAI as follows:

Generative artificial intelligence (genAI) is a new class of AI systems that can generate original content in response to user prompts. This includes text, images, audio, video, computer code, and data analyses. Unlike traditional AI systems, which are mainly designed to classify information, recognize patterns, or make predictions, genAI produces new material that resembles human-created content.

They go on to differentiate between two types of GenAI tools:

• General purpose (off-the-shelf GenAI tools)
• More specialised (or fine-tuned) for educational purposes (learning oriented GenAI tools)

Most of the GenAI tools that we are seeing in use in schools in Ireland are general purpose or off-the-shelf tools such as ChatGPT, Copilot, Gemini, and Claude. The research found, unsurprisingly, that learning oriented GenAI tools are more impactful on learning, and that the off-the-shelf tools bring distinct challenges, such as a prioritisation of financial gains over educational and equitable outcomes. However, as yet such tools are not available for Irish schools.

What are GenAI tools being used for by teachers – the reality

Our research is finding that most teachers are using GenAI tools to enhance their professional productivity by using these tools to create content that they then use with their learners. There appears to be limited use in the classroom with learners and to grade student work, and this may relate to the fact that most are using off-the-shelf tools. Yet, many teachers find that the use of GenAI tools can save them real time in creating content that is appropriate and differentiated for their classrooms. Similarly, many school leaders are also reporting that GenAI tools are helping them complete their work more productively.

Interestingly in Estonia, where teachers and students have access to learning oriented GenAI tools, namely ChatGPT for education, there appears to be limited classroom use by teachers and students. The Estonian Ministry has created professional development supports to help teachers consider how they might embed GenAI into their practice, so that they are using it ‘smartly’ with students. They are taking a holistic approach to GenAI use in their schools, one that considers the risks, while also considering the promise it has. Time will tell if it has a positive impact on the education system and in particular on student learning.

What might these tools be used for – the promise

While the Friend or Foe report stated, that GenAI “is a nascent technology, so research on its impacts on education is in its infancy and as yet inconclusive”, others are speculating that AI will ultimately transform how learning is organised. The Council of the European Union in their Draft Council conclusions on teachers in the era of artificial intelligence (AI) wrote that:

AI has the potential to influence how learning is organised, how teachers design, deliver and assess instruction, and how learners access and interpret information, and therefore may transform the relationship between teachers and learners; it may also affect how schools manage administrative tasks;

The Council highlights that AI has the potential to enhance:

1. active and collaborative learning and creative engagement;
2. inclusion, through the provision of educational resources for young people with disabilities and special educational needs;
3. accessibility for learners in rural and remote areas and from underrepresented and marginalised backgrounds;
4. participation of learners from diverse linguistic backgrounds in learning, thus enhancing cultural and linguistic diversity;
5. pedagogy through a variety of approaches, including practical and experiential learning;
6. personalisation to address diverse learning profiles, including individualised instructions and immediate feedback, thus supporting the acquisition of knowledge and skills;
7. assessment practices that focus on learners’ individual needs and progress, enabling early gaps identification and formative feedback and preventing risks of disengagement or dropout;
8. administrative efficiency, allowing more time to be dedicated to the process of teaching and learning

However, the evidence of such practices is, as yet, limited.

Some New Research from the US

While this is a new area of research evidence is starting to emerge and there AI Hub for Education in Stanford is one institution that is consistently sharing some interesting findings. Their recent report, The Evidence Base on AI in K-12: A 2026 Review, shared the following key findings:

Students

• Immediate gains with access: AI tools significantly improve student performance on math practice, programming projects, and writing tasks while students have active access to the technology.
• Short-term boost, uncertain transfer: AI improves performance with access but when assessed independently without AI support, effects are mixed.
• Easier doesn’t mean better: AI tools can alleviate students’ cognitive burden and foster positive experiences in learning but can be at the expense of deeper thinking.
• Pedagogical design matters: Tools designed with pedagogical guardrails (such as AI chatbots for tutoring that provide step-by-step reasoning instead of direct answers) show more promise than general purpose AI tools.

Educators

• Improved efficiency: Teachers using AI tools for lesson preparation spent less time on planning without reducing lesson quality.
• Scaling expertise: AI tools that provide regular, automated feedback and diagnostics to human tutors can improve instructional quality and student outcomes. AI pedagogical support can be particularly effective for less experienced and lower-rated tutors.

Their work is advocating that we take a learning science lens approach to understand how AI/GenAI could help or hurt students. They have selected six learning principles and again articulated the opportunity and risk associated with each.

Conclusion

To date most of the promise for the use of GenAI tools appears to be in relation to enabling teachers and school leaders to work more efficiently and faster, some might even say ‘smarter’. There is limited use of these tools, particularly in Irish classrooms, as yet but we can only assume that will change as developers create tools that are more specialised (or fine-tuned) for educational purposes. In the meantime, there is a need for greater research on the opportunities and risks of using GenAI tools in classrooms with learners, as we know young people are using these tools outside of school. There is a need to support teachers to be confident and competent in using these tools themselves, so they can decide:

• If they and their students should use GenAI
• If yes, what tools they might use and why
• How they design and activate learning activities, that use these tools appropriately
• If the use of these tools had an impact (they might even ask the students)

We have been here before with older technologies and we need to learn from these experiences and continue to explore how technologies, such as GenAI, can ‘enhance’ or even ‘transform’ our educational practices. Where we find they add value, we should share these widely with all educators. Conversely, we need to document where they don’t add value and share these lessons as well. We clearly have a ways to go on this journey yet.