Every year, Bett works with educators across the UK to better understand how AI is shaping teaching, learning and school leadership. Through our annual AI in Education reports, conducted with YouGov surveys, we explore not only how adoption is changing, but how attitudes, concerns and opportunities are evolving across the sector.
With our 2026 report launching later this year, now feels like the right moment to reflect on how far the conversation has moved since our first AI report in 2024.
In the last few years, AI has shifted from a growing curiosity to a daily reality for many educators. More schools are adopting AI tools, more teachers are experimenting with them, and conversations around workload, confidence and classroom impact have become impossible to ignore.
But while adoption has accelerated, the emotional and professional relationship teachers have with AI has also become more complex.
In 2024, more than two thirds of UK schools had not implemented AI. A year later, that figure had dropped below half. Across the same twelve months, the proportion of teachers using AI at least monthly climbed from a third to almost half. By almost any measure, AI has moved from the margins of the staffroom conversation to the centre of it.
But headline adoption figures only tell part of the story. Bett’s two annual reports, The Rise of AI in Education (2024) and AI in Education 2025: Navigating Progress, Pedagogy and Pain Points, reveal a profession that has not simply become more comfortable with AI. Its relationship with the technology has become far more nuanced.
This year-on-year analysis explores what has changed, what concerns remain, and what school leaders should take from a profession navigating one of the biggest shifts education has seen in decades.
The training deficit is growing
In 2024, the training picture was already concerning. A third of teachers in schools that had adopted AI reported receiving no training at all. By 2025, that figure had worsened significantly, with nearly half of all teachers saying they had received no AI support or training from their school whatsoever.
Yet when teachers do receive training, the impact is clear. Almost half of those trained in 2024 found it genuinely helpful, suggesting the issue is not that training is ineffective, but that too few teachers are receiving it.
What teachers are asking for is practical and achievable. They want hands-on sessions rather than theory-heavy presentations, and they want to learn from peers in schools like theirs who are already using AI successfully. More than a third also want clear school policies around appropriate AI use, not because they want to be micromanaged, but because uncertainty itself creates anxiety.
As one contributor to the 2025 report explained, a simple two-hour session combining practical tips, case studies and an overview of the tools can often make a meaningful difference. The barrier is not necessarily cost, it’s prioritisation.
The 2025 report also highlights a growing equity challenge. Research from the Sutton Trust found that independent school teachers are more than twice as likely to have received formal AI training compared to teachers in state schools. Without intervention, AI risks deepening the very inequalities education is working to close.
Workload remains the clearest win
Across both reports, workload reduction remains the most consistently positive outcome linked to AI adoption.
The proportion of teachers reporting that AI has reduced their workload has remained steady at around a third. At the same time, the number saying AI has increased their workload has fallen sharply, from nearly one in ten in 2024 to just one in thirty in 2025. AI is becoming less burdensome, not more.
The examples behind the data help explain why. The Education Endowment Foundation found that ChatGPT can reduce lesson planning time by almost a third, equating to roughly 25 minutes per week. Teachers in schools already using AI have described reducing a two-hour weekly newsletter task to 15 minutes, or cutting a three-hour risk assessment process down to just ten minutes.
For a profession already under pressure, these are not small efficiencies. They represent meaningful breathing room.
The emotional experience of teachers already using AI reinforces this point. Many report feeling less stressed, more empowered and more confident in their work. In practice, the experience of using AI tends to be more positive than the anticipation of it.
Which makes the training gap even more pressing.
The teachers receiving the least support are often the ones missing out on the benefits altogether, and are therefore more likely to remain anxious about technology they have not yet had the opportunity to explore safely.
The emergence of a professional identity crisis
One of the most striking findings in the 2025 report is the emergence of what it describes as a “professional identity crisis” around AI use.
Almost half of teachers feel they are “cheating” when they use AI for core teaching tasks, while an identical proportion believe they are not doing their job properly when they rely on it. These are not fringe opinions. They point to a deeper tension between what teachers have traditionally understood their role to be, and what it means to delegate parts of that role to technology.
In 2024, the dominant concerns centred on AI’s capabilities. Could it detect cheating? Could it help address teacher shortages? Would it ever become reliable enough to trust?
By 2025, the questions had shifted inward. Teachers are no longer primarily asking what AI can do. They are asking what AI means for them as professionals.
That response is entirely understandable. Teaching has always been seen as a deeply human profession built on care, judgement, expertise and relationships. A tool capable of generating lesson plans, drafting reports and providing feedback naturally challenges long-held assumptions about where professional value lies.
There is also a growing shift happening within the classroom itself. Nearly a quarter of teachers now feel less confident using AI than their students, while almost a third feel actively intimidated by their students’ greater AI knowledge. The traditional classroom dynamic, where the teacher is assumed to be the expert in the room, is becoming more complex.
This cannot simply be dismissed as resistance to change. School leaders need to acknowledge these concerns openly and help teachers build a professional identity that positions AI as a tool they direct and shape, rather than something that diminishes their expertise.
What teachers still believe AI cannot do
Amid all this change, one thing has remained remarkably consistent. Teachers remain clear about where AI stops and their own irreplaceable role begins.
Across both years, teachers were emphatic that the most human aspects of education cannot be replicated by technology. Supporting children through relationships, failure and responsibility. Helping a struggling student feel genuinely seen. Creating the classroom energy that makes learning memorable.These are not tasks teachers believe can be automated.
Around four in five teachers believe AI will never replace the teaching of essential life skills, while similarly large majorities feel the same about hands-on learning and genuinely personalised support. This is not resistance to innovation, it is a grounded understanding of what great teaching truly involves.
As the 2025 report puts it, the future is not a battle between humans and machines. It is teachers working alongside intelligent tools to achieve better outcomes. The real risk is not that teachers think too little of AI, but that anxiety, unclear policies and inadequate training prevent them from benefiting from what it can genuinely offer.
Student concerns are becoming harder to ignore
The 2025 report also introduces a much fuller picture of how teachers feel about student AI use, and the findings are more concerning.
Plagiarism remains the dominant issue, with roughly seven in ten teachers worried about it. But the concerns go much deeper than academic misconduct. A majority are worried that students may absorb inaccurate AI-generated information as fact, while more than half fear students are losing opportunities to develop important skills through practice and independent thinking.
At the same time, the picture is not entirely negative. Around a quarter of teachers are genuinely excited by what students can achieve with AI tools, showing that attitudes remain mixed rather than uniformly pessimistic.
The academic integrity debate has also evolved significantly since 2024. Initially, concerns focused largely on whether AI would make cheating more difficult to detect. Now, the conversation has become broader and more fundamental. Is AI beginning to change what students actually learn?
Research referenced in the 2024 report found that GPT-4 significantly improved student performance on practice problems in maths classes. However, when access to GPT-4 was removed, students performed worse in subsequent exams compared to those who had never used the tool. It is a powerful reminder that AI can support learning in the moment, but without careful use, it may also weaken the development of genuine understanding.
This does not mean restricting access to AI altogether. It means thinking far more carefully about how learning and assessment are designed, ensuring AI becomes a tool for deeper thinking rather than a shortcut around it.
What school leaders should take from this
Two thirds of teachers expect to increase their AI use over the next twelve months. That represents a remarkable shift for a profession that, just a year ago, was still largely observing AI from the sidelines.
But momentum alone will not be enough.
Teachers are not asking for lengthy lectures on AI ethics or technical deep dives. They want practical opportunities to test tools, experiment safely and learn from peers facing similar challenges. A well-designed two-hour session can significantly shift confidence levels without requiring major investment.
Many teachers are also asking for clearer guidance around appropriate AI use. Not because they lack professional judgement, but because uncertainty leaves them second-guessing themselves. That uncertainty directly contributes to the feeling of “cheating” that so many teachers report. Clear boundaries reduce anxiety, and anxious teachers are far less likely to experiment confidently.
The professional identity crisis highlighted in the 2025 report is real, and ignoring it will not make it disappear. School leaders who openly acknowledge these concerns, and who help staff understand that thoughtful AI use is an extension of professional expertise rather than a replacement for it, will be far more successful in driving sustainable adoption.
The widening gap between independent and state school AI training also deserves urgent attention. Schools with limited budgets should recognise that impactful training does not have to be expensive, and that successful approaches can and should be shared more widely across the sector.
Taken together, Bett’s 2024 and 2025 AI reports chart the evolution of a profession genuinely grappling with transformational change.
In 2024, scepticism dominated the conversation. Teachers questioned AI’s reliability, trustworthiness and place within education.
By 2025, that scepticism had evolved into something more complex. Teachers increasingly recognise that AI is here to stay, but many are still uncertain about what that means for their identity, confidence and role as professionals.
In many ways, that shift is a sign of progress.It shows that teachers are taking AI seriously.
The challenge now is whether school leaders, policymakers and the wider education sector are prepared to take teachers’ concerns just as seriously in return.
The tools already exist. The evidence of benefit continues to grow. What remains essential is investment in training, time and trust, so every teacher has the opportunity to explore AI confidently and thoughtfully, not just the early adopters already leading the way.
Sources: The Rise of AI in Education 2024 and AI in Education 2025: Navigating Progress, Pedagogy and Pain Points, published by Bett and based on YouGov surveys of UK teachers.
