AI Education Content Store

MONTHLY NEWSLETTER #3 - MARCH 2025

This newsletter aims to keep you updated on the development of the AI Education Content Store, commissioned by the Department for Education (DfE) and the Department for Science, Innovation and Technology (DSIT).

In this edition:

What is the AI Education Content Store?

Can you help us: We’d love to hear from interested schools and teachers

The Project in development: 

Meet the Team: 

Who are we working with? 

Coming Up: Dates for the diary

What is the AI Education Content Store?

The AI Education Content Store will be a repository of AI-optimised educational material, developed in partnership with the sector and key education organisations. It will include curated content such as programmes of study, lesson plans and anonymised student work.

The material in the content store will be gathered collaboratively with the sector – we will be guided by teachers, students and parents, as well as by educational leaders. Once we have amassed a centralised repository of AI-optimised content, we will offer access to edtech providers, who will use it to develop AI products for schools and colleges.

Can you help us?

We are working with schools to gather content for the AI Education Content Store, and we’re very keen for all interested schools to be able to participate.

If you’d like to be involved, please email: content-store-programme@faculty.ai

The Project in Development:

The second AI Education Content Store hackathon was held over two days at the end of March.

Attendees included teachers, school leaders and multi-academy trust leaders, drawn from our teacher and MAT reference groups, as well as Department for Education AI for Education tools’ competition winners and other edtech companies that have shown an interest in the project. Content providers, such as the Education Endowment Foundation, were also represented.

Education Secretary Bridget Phillipson and Peter Kyle, Secretary of State for Science, Innovation and Technology, attended the first day of the hackathon. They were given a brief introduction to the content store and met three edtech companies. These companies gave them a demonstration of tools they’re currently developing, and explained how the content store could be used to enhance them.

How did participants use the AI Education Content Store?

After general and technical introductions to the content store, attendees split into groups to come up with solutions to a series of problem statements. These problem statements were each designed to tackle one of four use cases:

Generating practice summative assessments

Generating targeted formative assessments at a given point in a scheme of learning

Marking assessments (long-form or subjective)

Providing targeted feedback in line with best practice

So, for example, groups looking at generating summative assessments tackled one of the following problem statements:

         The challenge is to create a tool which can generate practice SAT questions to be used as part of lesson plenaries. The tool should be able to generate questions for students working at a range of attainment levels and reinforce the concepts covered in the lesson delivered. The tool should also generate marking guidance which mimics that typically used for real-world SATs.

         The challenge is to develop a platform which creates differentiated practice SATs assessments for cohorts of primary students. The tool should be able to ingest data about students and batch-produce individualised practice SATs which are shaped by individual student needs/areas for improvement. The tool should also generate marking guidance for teachers which mimics that typically used for real-world SATs.

And the groups looking at providing targeted feedback addressed the following problem statements:

        The challenge is to build a system that analyses schoolwork by key stage 1 or 2 pupils and generates personalised, best-practice-informed feedback. The tool should also be able to generate multiple versions of the same feedback for a teacher to pick from (different tones, different reading ages, longer, more concise, etc).

         The challenge is to create a platform able to provide targeted feedback on key stage 1 or 2 schoolwork. The tool should not only be able to generate feedback but also surface the guidance referred to in the creation of the feedback, supporting teachers with the development of their own practice.  

Groups used ChatGPT and other large language models to examine how “base” LLMs would respond to their problem statement – creating a baseline for comparison. After that, they linked the AI Education Content Store to LLMs as part of their solution development, supporting a comparative analysis between LLM responses that were and were not backed by the content store.

For example, the content store now contains marked key stage 2 essays and EEF guidance on how to give feedback, both of which the tool would be able to draw on in order to generate essay feedback. Users could then adjust the structured feedback according to the tone that children responded to best – requesting, for example, that it should be written in a humorous way.

Teams were also provided with dummy pupil data, allowing them to create tools that differentiated for the fictitious children in the group.

What was new for this hackathon?

On the second day of the hackathon, the teachers, school leaders and MAT representatives swapped teams, acting as product testers for other teams’ solutions. They offered feedback on why the suggested product might be useful – or otherwise – for them.

At the end of the second day, each team gave a presentation about the product they’d designed. They explained which content they’d used from the store and itemised any additional content that they would have found useful.

One group's content store-backed prototype was linked to LaTeX, a document-preparation system that can generate scientific notation and diagrams. This allowed hackathon participants to generate sample maths questions for key stage 2 summative assessment.

Participants reported that the additional content that had been added to the store since the January hackathon significantly increased the quality and precision of the output they were able to produce. It also increased the potential use of the content store to them, and boosted their confidence in the programme.

In addition, a significant proportion of the edtech companies attending the hackathon use AI to help with assessment and feedback, so the focus of this hackathon was directly relevant to them.

Edtech CEOs and CTOs in particular were excited that they had been able to build a product in two days that, within two to three months, could be taken out to market – it gave them a very tangible sense of how the AI Education Content Store could potentially expedite their development processes.

Meet the Team:

Tom Nixon

What is your role? 

I’m managing director of Faculty’s ‘Applied AI’ business, and I particularly oversee our work in education. 

I oversee wider delivery of the AI Education Content Store, making sure the relevant streams are all on track. I have strategic oversight of the project, ensuring that all the different elements fit together and that the whole programme is making sense. 

I’m also working with broader stakeholder engagement across the sector. For example, I’m currently having conversations with key leaders in the further-education sector about how the content store could better support vocational and technical education in FE. And I’m helping to engage exam boards, and discussing what their involvement might be in the content store. 

What’s your background?

I’m an economist by original training. I worked in government for 10 years, across a variety of areas, including education, technology policy, healthcare and business policy. I also worked in Downing Street, particularly focusing on technical education and apprenticeships.

Because I’m an economic analyst by training, I could see the potential for data science and data analysis to solve public-policy problems in new and better ways. That’s very exciting – and led me to join Faculty AI in summer 2019.

What do you enjoy about working on the AI Education Content Store?

I enjoy the engagement across the sector. We need to speak to so many different groups of people: school leaders, teachers, edtech companies. That breadth of voice is really exciting and refreshing.

It’s also bringing us into contact with fantastic edtech companies doing really exciting things. For example, one of the companies that attended the January hackathon is using voice recognition and transcription to allow nursery staff to record progress against early learning goals. They want to build a fantastic product, but they need it to be aligned to the relevant curriculum. They’re not going to want to spend time processing Early Years Foundation Stage guidance and all the documentation around that. So that’s where we see the real ability of the content store to enable and support all this great stuff happening in the edtech sector. 

What do you do when you’re not working on the content store?

I have two boys, aged seven and three, so most of my spare time is spent on family stuff – school drop offs and taking them swimming. I cycle to and from work, and I’ve been teaching the boys to cycle. 

Who are we working with?

We have now held several rounds of discussions with of our edtech reference group. During these sessions, we demonstrated the AI Education Content Store, with a specific focus on the API, which is a set of instructions that allows different software programs to communicate with the content store.

We asked the edtech companies for feedback: how would they want to use and access the data in the content store? Does the structure of the API serve their needs? So far, responses have been positive.

We are in the process of conducting one-to-one interviews with group members for more detailed feedback. In particular, we’re discussing with them the kind of content they’re looking for, and how they’d like it to be presented.

Feedback so far has been that the data model makes sense, in terms of tagging content by year group, key stage and topic. Some have also made suggestions about additional tagging that could be useful to them.

And edtech companies are interested in the kinds of content that we’re already in discussions about acquiring for the content store.

They were also interested in understanding the content store’s data modelling, so that they can ingest content directly from the store, fully optimised.

For example, while their own systems already house a copy of the national curriculum, it could be useful to them to adapt how they ingest it, to fit with the content store’s modelling. This would mean that when the national curriculum was reviewed and updated, they would be able to pull in the new version from the content store – fully labelled and tagged – rather than updating it themselves manually.

Coming Up...

April: Ongoing meetings of multi-academy trust, college, school, edtech and content-provider reference groups.

Got a Question for the Core Delivery Team?

email us at: content-store-programme@faculty.ai