Next steps for AI in education

Framing the day, Professor Miles Berry, Computing Education, University of Roehampton, encouraged delegates to look at the broader scope of what education really means given that much of AI use by young people happens outside ‘the walled garden of school’.“Many of us are particularly concerned about the dangers of cognitive outsourcing and cognitive offloading – of having the machine do the thinking for us and the thinking for our young people and that raises huge challenges,” said Berry.Along with current concerns, Berry highlighted interesting AI developments that go beyond lesson planning or assessment, including an AI tutoring trial for disadvantaged pupils. He urged delegates to explore both the medium and long-term implications of generative AI in the face of a changing employment market.

AI use in schools

First up was Oliver Large, policy advisor, Tony Blair Institute for Global Change (TBI) who ran through the current use of AI in schools, emerging challenges and his recommendations to scale safe and high quality AI in schools.“AI is already in our schools. It's not in some distant scenario, or in a pilot at the edge, it is in pupils' pockets and in their homes, the question, therefore, and what we focus on at TBI, is whether we shape that change deliberatively or whether we let it happen by accident, unevenly, inequitably, and poorly.”Large took audiences through some TBI research. Going through the headlines of their polling, he stated that pupils were not waiting for permission to use generative AI.“By the end of 2025, 80% of secondary school pupils were using generative AI tools for their schoolwork, but this is overwhelmingly outside of the classroom, so the pattern here is clear. Lots of informal use, but very little structured use.”On teacher adoption, he added that while around two fifths of teachers are using generative AI at least sometimes, where teachers are using it, it is often around a very narrow set of admin tasks like generating lesson materials or curriculum planning which is useful, but only a part of what AI could do for educators.In terms of school level adoption, Large implied the data is more stark. With only around one fifth of primary and secondary schools having a policy in place on the safe and appropriate use of AI. While only one in five teachers say anyone in their school is teaching pupils what AI is and how it works.

An emerging divide

Furthermore, he noted that who gets to develop the AI skills needed to thrive is not evenly distributed. With independent schools being "almost three times more likely than state schools to have a school-wide AI strategy, more likely to teach pupils how to use AI, and to teach how AI applies within subject areas.”The TBI report warns an emerging social divide could create a new divide on top of the old one. A divide that is not just about access to technology; but access to opportunity, confidence and the AI skills necessary that lead to success in the future.“This is fundamentally a progressive opportunity for everyone in the education space. AI proficiency and ensuring a safe and effective supply of AI products cannot be left to better resourced schools, or a small group of pupils. It must be a universal entitlement, and the choice for English education systems is whether it leads or follows,” said Large.With informal student use of AI, shallow teacher use, a limited whole school strategy and a new digital divide that is opening up in real time, Large took a closer look at what is driving that.

---

Read related articles

• How much AI is too much?
• Taking a child-first approach to AI literacy
• Building students for an AI-driven workforce
• AI literacy and competency: The key to workforce readiness

---

Obstacles ahead

Large shared five areas the report identifies are within the government's power to fix in order to get young people AI ready.“The first is leadership. School leaders are waiting for clear signals, they're cautious, understandably so, given the risks involved. There is also a lack of teacher confidence in using AI. Our polling highlighted that 91% of those who use AI are entirely self-taught and I think that links to a kind of a systematic training pipeline problem.”The third, noted Large, is information asymmetry as schools struggle to get the right information to tell good products from bad ones, with teachers left to navigate a noisy market. The fourth is digital infrastructure – a continuing problem around the digital divide.“Our polling highlights that barely half of secondary schools have reliable whole school Wi-Fi, and around the same figure have access to laptops for in-class learning. You can't run AI-enabled learning on patchy Wi-Fi and inadequate supply of devices. The final driver is market dysfunction.”The market dysfunction the TBI research refers to is a significant reduction of investment in school technology which Large stated has fallen by around 96% since 2020 with just 0.7% of ed tech investment going into AI tools. By contrast, the equivalent figure for AI investment in the health tech sector is around 22%. Implying, said Large, that the market is not delivering what schools need and capital is not flowing in.The accelerated pace of AI adoption, AI literacy and the need to have it deeper embedded in the curriculum to become more of a core outcome of schooling was discussed too.

Cognitive offloading

A key topic was cognitive offloading – where students use AI to bypass learning rather than deepen it, avoiding the building of a knowledge foundation and the development of skills. An area discussed by many on the day and listed by the TBI as an emerging challenge along with a lack of vision on AI leading to a ‘strategic drift’. However, solutions to combat cognitive offloading are far from simple, with suggestions by some that a more multi-faceted policy approach would work better than an enforced sanction or ban of AI use in classrooms.Sally Thomas, senior policy manager at the National Education Union (NEU) which brings together the voices of nearly half a million educators and school leaders, talked about some of the challenges their members are facing.‘There's a real necessity to ensure that AI is being used in a way that really supports learning without harming cognitive development, as well as social and emotional development and wellbeing, and this is something our members are really concerned about,” she said.“In an NEU survey asking if AI use amongst students had declined their critical thinking skills – 66% of secondary teacher members said that they thought it had and we know this is something that is backed up by wider evidence and research so we really need to address those risks.”

AI and assessments

Speakers went on to explore how teachers and students are using AI tools in different ways from learning to assessment.While others discussed ways schools are able to automate and integrate AI into LMS systems for more adaptive and personalised learning for students. As well as being able to generate reports, manage timetables, track attendance and flag concerns. Delegates shared how the use of AI marking tools are becoming more widespread with the growing accuracy of tools such as No More Marking.Meredith Reeve, a research strategist at Pearson, shared findings from a report on the impact of generative AI on formative assessment and teacher’s perception of this in light of extensive media coverage over the last year on cheating.Interestingly, she revealed a perception gap when it came to AI and formative assessments.“When we asked teachers whether they thought their own students were using generative AI for formative assessment, only 24% did, which is at odds very much with self-reported data by students across different surveys over the last couple of years, which shows students at the region of 80-90% use and growing steadily. What was also striking, was that only half of teachers reported that their school or college had an AI policy in place.”The Pearson School and College report  found that just 16% of school teachers and 27% of college tutors say they feel confident in teaching students about AI. Raising questions about what guardrails are in place and the confidence that teachers might feel to innovate safely, said Reeve, echoing others on the day.Looking forward, she noted a number of ways to improve formative assessment design for the current speed of change and urged delegates to think of it as more of a ‘basket of evidence’ demonstrating what students know and can do. Recommending an assessment approach that is more holistic, student-centred, tighter aligned to learning outcomes, and multimodal.“We know the speed of change here is unlike anything we've seen before. We're talking about change in weeks and months rather than years. We also know that education at system level is a slow-moving ship. So, this is a challenge, but by allowing students and teachers the permission to innovate within carefully designed guardrails, there's an opportunity for us to strengthen formative assessment practice, and to make it more effective, whilst also developing the AI literacy and critical thinking skills that are going to be so important for students in their work and in their lives alongside AI.”Westminster Forum disclaimer: Please note speakers have not had the opportunity for corrections

---

---

Find out more about the Think Global People and Think Women community and events.

---

Subscribe to Relocate Extra, our monthly newsletter, to get all the latest international assignments and global mobility news.Relocate’s new Global Mobility Toolkit provides free information, practical advice and support for HR, global mobility managers and global teams operating overseas.

©2026 Re:locate magazine, published by Profile Locations, Spray Hill, Hastings Road, Lamberhurst, Kent TN3 8JB. All rights reserved. This publication (or any part thereof) may not be reproduced in any form without the prior written permission of Profile Locations. Profile Locations accepts no liability for the accuracy of the contents or any opinions expressed herein.