Learning in an AI-shaped world

Experts across academia and industry discussed the next steps for AI in education, covering learning to research, policy and teacher capability.Themes included cognitive offloading and outsourcing, equity and inclusion, as well as a lack of policy and centralised guidance on AI use in education. There were calls too for a national ed tech strategy and more independent research on how AI tools impact various outcomes and teacher workload.The potential of 1-2-1 AI tutoring was explored with some concern that students could miss out on a vital benefit teachers provide – getting young people motivated and interested. While many expressed the need to keep teachers at the heart of decision making when it comes to AI and ed tech in learning.

On AI use and misuse

Daniel Wilson, Senior Product Manager, Data, AI, Skills and Social Responsibility, UK Impact, Amazon UK: AI is like a whetstone for our abilities. If it's used well, it sharpens your skills, but if it's used carelessly, it may dull your edge. And whilst access matters, worse than no AI skills might be bad AI habits, especially in the young, because they don't have a pre-AI playbook to fall back on. So, a simple test that I like to challenge people with is: if AI disappeared tomorrow, would you be better off having used it? What have you gained that you can carry forward? As AI can handle more and more tasks – we need to make ‘effort’ cool.We need to call out the sloppy AI use. When we showcase it to young people we need to engineer ways where humans are clearly aided by using the AI but we also need to showcase times where they're clearly hindered by using it. We need to frame its outputs as opinions and not facts, and we need to encourage people to pick AI's role to suit the task that they've got in front of them. Should you treat it as a tutor to teach you something? Should you treat it as a partner so it can challenge your assumptions? Or in some cases, but never all, to outsource your work to it?Some put off AI learning because they perceive it to be too technical a thing to even get involved with, but with the right support I've seen for myself how people grow in confidence. The technical depth that we need to teach people around AI can be an obstacle to further engagement so the timing of it is really important. We also shouldn't assume that more classically technical profiles of people will be the most proficient with generative AI – and that can help widen the pipeline for more technical talent in the future.Prof Miles Berry, Computing Education, School of Education, University of Roehampton: On cognitive outsourcing the research is pretty well documented. We have a strong evidence base that the AIs are really very good and produce super work in all sorts of territory but they don’t necessarily help anyone to learn anything. The thing that you get good at is the thing you spend your time on, and that thing could be just prompting the AI, or it could be learning the subject.Dr Ajaz Ali, Director, Higher Education, BMet College: I think what we don't discuss enough, which is the elephant in the room, is the AI hallucination and error rate – where it generates something with great confidence and will -give you proper references and everything but the information itself will be completely wrong and absolutely fabricated which is quite misleading. The percentage of fabrication or hallucination with AI can be up to 70% at times.Gemma Gwilliam, Head of Digital Learning, Education and Innovation, Portsmouth Digital City Project: In the world of assessments we've heard about AI wearables. I know that Digital Futures Group and I put out guidance for schools across the UK on AI wearables but are we checking and seeing whether or not these are being used in exams? Can schools recognise whether or not someone has walked into an exam room with a pair of AI glasses on, where they could have the notes already stored in their glasses? Are we fully aware of what's available? Because I know that our children are, and I know that they can buy them off TikTok shop for as little as £8.82.

---

Read related articles

• Next steps for AI in education
• How much AI is too much?
• Taking a child-first approach to AI literacy

---

On AI literacy and work readiness

Sally Thomas, Senior Policy Manager, National Education Union: We need AI literacy to be explicitly embedded within the wider digital literacy curriculum. It’s really critical in terms of increasing student use of AI but also to understand AI's capabilities, limitations, and wider social implications. In terms of how AI should be used, we really want it to be inquiry-led and support deep learning, so it's really mitigating those risks we know AI can have around shortcutting thinking and cognitive development.Olly Offord, Policy Adviser, Science Education and Skills, The Royal Society: If AI literacy is going to be covered by a broad category (not just taught in computer science or computing) and the responsibility of all teachers, so they act like a nexus point for young people's interaction with AI in formal settings, then we really need teachers to feel confident about AI. I'm afraid to say they're not.Benjamin Barker, Principal, King’s Leadership Academy Wavertree and Director, AI, Great Schools Trust: It’s a bit of an uncomfortable truth, but AI is probably going to reward people who can think clearly, communicate precisely, evaluate the information they're receiving critically, and adapt fast as well. Those that can't risk becoming maybe some of the more economically vulnerable moving forwards, so as a school we have a major responsibility.Dr Ajaz Ali: Students can identify the answers much quicker than the tutor is teaching them so that is a challenge. I think that can only be replaced by having more vocational, hands-on experiential learning, rather than conventional teaching styles. Workforce requirements are changing too. There is a massive gap between education and the expectations of industry, a shortage of skilled graduates and everyone needs continuous reskilling. I think the teaching workforce needs more reskilling than any other profession at this stage.Prof Miles Berry: We're nearing the 10th anniversary of Brexit. If it hadn't been for Brexit, we would have had to implement AI training for everybody working in an organisation that is using any AI system. Thanks to Brexit, what we have is the provision of content without any compulsion to use it.

On the future of assessment

Dr Madhia Khan, Director, Research and Consultancy, EDUCATE Ventures: I've really been struck by a range of different stakeholder engagements that we're doing and it’s this tension between what school and college leaders are thinking about versus students. With school and college leaders, the worry is very simple. They worry that students are using AI to skip the thinking, that there's an erosion of their cognitive abilities, and they're getting the answer without doing the work. And often, the way they respond to that is to put in place firmer boundaries and police AI in different ways.From students, it's the opposite, they're often worried about being accused of using AI when they haven't. So actually, we get situations where students are dumbing their work down deliberately because they're worried that if they present a piece of work that shows progress they'll be accused of having used AI when they haven't. They take out their best words, they put spelling mistakes in, and they in some ways are working within the system they've been given and are inhibiting their own learning as a result.There's an interesting phrase and piece of research done by WONKHE about being ‘trained to stop learning’, and I think that is something which is inadvertently happening here. There are many students who are starting to worry that their grades are not reflective of what they actually know themselves. Four in 10 are admitting handing in work that they can't explain, and only some feel that their assessment is genuinely connected to what they are learning.On the other side, we are also seeing some very interesting practices on the ground of how AI is being used within formative assessment.Gemma Gwilliam: Assessment no longer reflects how learning happens. Traditional assessment models are based on individual unaided work, but for me, this isn't about detecting AI use, it's about recognising that supported learning is now the new normal, and providing our pupils with those digital tools isn't anything new. AI is just that extra tool in the toolkit that we are monitoring, supporting, and teaching people how to use safely.

Did you know? South Korea has announced it is training every teacher to use digital tools by 2026. While Singapore aims to roll out AI tools to every classroom by 2030.In 2025, Estonia launched AI Leap – which provides AI-powered learning tools and generative AI models to teachers and pupils, with half of its schools employing ed tech lead teachers who run on-site labs and train colleagues to integrate technology effectively into their teaching.Source: Tony Blair Institute for Global Change

_{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.