Beyond the upgrade: a theoretical framework for education in the age of AI

There’s nothing quite like thinking about grading in the age of AI to make you realise how quickly the ground under our feet is shifting.

Right now, teachers are using AI to grade students’ work. At the same time, students are using AI to produce work to be graded. You don’t have to think too long before you see the loop forming, AI grading AI-generated work, until possibly the very idea of grading starts to eat itself. 

We’ve heard a couple of stories about how to get over this. In the first, a school that’s adapted by asking students to “show their workings” again. This reminded me of long division when I was at school and all my workings would count, not just getting to the right answer. Here, the prompts would be graded alongside the output. 

In another classroom, the teacher asked students to use AI at home to draft an essay on a topic, then bring it in the next day. In class, they compared versions, spotting where the answer differed and thinking critically, together about how to surface truth and facts from what had been generated. In other words, they used AI as a starting point for critical thinking, not the end point. And instead of marking logical thinking as part of the process, they created a classroom that enabled critical thought, dialogue and collaborative problem-solving in groups.

These stories matter because most of the conversation about how to use AI for education right now is about the upgrades to existing tasks and processes, in the assessment example the upgrade is using AI tools to assess work. But upgrades can disguise disruptions. If we only optimise the status quo, we risk optimising ourselves into irrelevance.

The AI Observatory has created a theoretical framework to help make sense of this change

Our role is to generate timely, practical evidence on strategies that can narrow the learning divide in the age of AI. We take a hypothesis-driven approach, testing and learning what works. The framework is designed to support intentional decisions about AI today, while keeping sight of where education needs to go next. It also helps us to prioritise which ideas we might test. 

The framework is grounded in two key insights: 

1. “AI in education” and “education in the age of AI” are not the same thing

We have organised decisions in line with three different horizons of change. 

The first is “AI for education” an upgrade to existing systems. In the example above it’s where grading is made easier using AI. 

The third horizon is “Education in the age of AI”, here we are educating and learning in an age powered by AI. In the example above, this is where pedagogical approaches adapt in response to a new educational context, just like in the grading example at the top of this piece

In between, we have the disruption, “AI in education”. The idea is that the more we can imagine the third horizon and design for that reality, the less disruption we will feel. The second horizon outlines what happens when the elements of our education systems no longer make sense. When exams, homework, and even the rhythms of classrooms themselves need reimagining and we begin to shift the status quo.

Our framework at EdTech Hub builds from this. It recognises that decision-makers today still need to focus on the upgrade, there are efficiency gains to be had, especially in ministries where AI can improve data use, speed up processes, and help make decisions in resource-constrained contexts. But the framework also asks how we prepare for the bigger shifts.

2. If we want real progress, there are many levers to pull

We’ve surfaced six levers where AI is already reshaping education: what students learn, how teachers teach, and how systems run. Together, these levers show that each of us has a role in making sure our actions fit the realities of the contexts we work in.

1. Enable learners – personalised, adaptive learning and future-ready skills.
2. Empower teachers – more time and headspace for teaching and supporting students.
3. Streamline bureaucracy – smarter planning, faster responses, smoother systems.
4. Align partnerships – private and public actors coordinated around learning outcomes.
5. Create context-driven solutions – tech shaped around local needs, not the other way round.
6. Renew the purpose of learning – rethinking what education is for in the AI age.

A theoretical framework: 18 ideas for narrowing the learning divide in the Age of AI

By mapping the three horizons of change against the six levers, we could see where different kinds of disruption and transformation might happen. 

At each intersection, we identified an idea for narrowing the learning divide in the age of AI and have 18 in total.  We have used this to prioritise our own work in the AI Observatory, and we’ll be capturing and sharing evidence as we go on the role these ideas might play in narrowing the learning divide. 

We hope this framework helps spark ideas and pushes us all to think beyond upgrading what we have. We’d love to hear your thoughts, and we’ll keep sharing what we’re learning as these ideas are tested in practice, in our work and beyond. Let us know what you’re seeing too.

---

We’d love to hear from you! What’s been shaping your thinking on AI? Drop your thoughts (and reading recommendations) in the comments on LinkedIn. Explore more from EdTech Hub’s AI Observatory.

EdTech Hub’s AI Observatory is made possible with the support of the UK’s Foreign, Commonwealth and Development Office.