Lesson 03
Inspired Education
A mountain of paperwork doesn’t draw anyone into teaching - but it certainly drives them out of it. At Inspired Education, AI is freeing teachers from repetitive tasks and allowing them to focus on what they love best: helping kids learn.
At 10am on Monday morning,
eleven year-old Millie Brown sits down in her classroom in west London, opens her laptop, and starts a test on the science she’s been learning. When she’s finished, she clicks ‘Submit’. When the test is returned to her it’s been marked, corrected and annotated with useful feedback to help her improve her answers next time - after being verified by her teacher, Mrs Roberts. In addition, Millie’s scores have already been uploaded into the school’s data system, so that Mrs Roberts can monitor her progress, and share it with Millie’s parents.
The whole process - creating the test, marking it, providing feedback and uploading the results - has been done by AI.
In the classroom next door, Fred Jorgenson is teaching his students about the history of the US Civil Rights Movement. On the electronic whiteboard, he takes his class through a series of slides he’s prepared. One shows annotated images of Martin Luther King leading the march on Selma. Another hosts a list of thoughtfully-designed activities to consolidate the lesson materials. The whole presentation is well designed and professionally presented - no corny clipart or clashing fonts - and designed to move at the right pace for his class, engaging the more able students while making sure everyone is able to keep up.
In his last job, Fred was sometimes up until past midnight preparing lesson plans and presentations for the next day. Now he’s well-rested, with more energy to spend inspiring the kids in his classroom. He’s still in charge of planning his lessons, making sure they’re right for his students. But the heavy lifting is done by AI.
Millie, Fred and Mrs Roberts are fictionalised characters who represent real people. But in most classrooms in the world right now, their experiences would be beyond imagination.
The paperwork challenge for modern teaching
Education resists change. Outside the school gates, pupils can access all the information in the world in the palm of their hand, and dream of jobs that didn’t even exist when they started formal education. Inside school, things are slower to evolve. A Victorian schoolteacher who dropped through a time warp into an average 21st century classroom wouldn’t have much trouble knowing what to do.
Nadim Nsouli is on a mission to change that. The Lebanese-British businessman is the founder and Chief Executive of Inspired Education, the world’s leading premium private school provider. From a standing start in 2013, Inspired now operates over a hundred schools across 24 countries, from Ho Chi Minh City to Rio de Janeiro. On any given school day, some 8,000 teachers are teaching over 90,000 pupils in its classrooms.
After a successful career in law, investment banking and private equity, Nadim entered the world of education with an ambition to take a fresh approach to schools, rethinking traditional approaches to the curriculum and pedagogy. In short, preparing students for 21st century life by teaching them with 21st century methods and tools.
Nadim explains, ‘The driving force behind the creation of Inspired Education was my vision to unite the world’s best schools under one banner, facilitating the sharing of global best practices while preserving each school’s unique identity and values. I took the leap from private equity to schools because I wanted to create a schools group designed to nurture well-rounded individuals who are not only academically accomplished but also confident and capable leaders. Our motto, “Embracing Individuality. Preparing Leaders,” reflects this holistic approach.’
At the outset, Nadim was a newcomer to the world of education in more ways than one. Not only was he at the helm of a growing company in the educational sector, but he was also experiencing the system from a parent's perspective, as his child entered school. This dual vantage point revealed a recurring issue that he sought to address: teachers across the sector were overwhelmed with responsibilities, and parents struggled to access straightforward information about their children's progress.
Because while the time-travelling Victorian schoolteacher might take the teaching in her stride, what would really blow her mind is how much time teachers now devote to paperwork. Outside the classroom, they spend countless hours creating materials and slides for each lesson, marking hundreds of pieces of homework per week, writing reports and responding to parents’ admin queries. For many, it crowds out all the benefits of the job. In the UK, for example, almost 20 percent of newly-trained teachers leave the profession within two years of qualifying, with many more considering quitting. Type ‘teacher retention’ into Google, and the first suggestion you get is ‘crisis’.
If Inspired were going to reimagine education, they first had to rethink teacher workload.
Enter the Inspired Global Study platform
In 2019, well before LLMs became mainstream and even before Covid introduced a generation of parents and children to online lessons, Inspired set about preparing the ground for the deployment of AI in their schools.
A key driver of this was Torben Lundberg, a straight-talking Dane with over 20 years’ experience in IT management, who joined Inspired as Chief Information Officer. The company developed a modern Azure-based data platform across their schools that linked all pupil, parent and teacher data relating to teaching, learning and school administration: from individual children’s grades and progress, to company-wide finance and HR systems. For a lot of sectors, putting all your data on a single platform is a no-brainer. For a group of schools, it was a world first.
In May 2021, Inspired acquired the online-only school InterHigh and combined it with its own virtual offering, King’s College Online. These pioneering schools allowed pupils to access a virtual education from anywhere in the world. Torben and his colleagues took that expertise and built it out into a proprietary online learning platform - the Inspired Global Study platform - that would work for students in any of its schools. It complements the world-class teaching in Inspired’s classrooms with the wraparound benefits of remote learning, so that pupils can easily catch up on missed lessons, access additional content and stretch material, and submit homework assignments. Crucially, it also helps parents support their children by giving them fine-grained information on what the kids are doing and how they’re getting on.
Testimonials
"I could see [AI] would be big in many areas of our company, but particularly the core academic processes. And the reason we’ve been able to move much faster in AI than other educational companies is because we had the platform. No-one had as much data as we did."
In fairness, none of this was completely revolutionary. All schools have management information systems, finance systems, HR systems and so forth; and most now have some kind of online learning, behaviour and parental engagement platforms, even if it’s just paying for school meals and assigning homework. The difference is that most schools buy in a suite of different products that don’t talk to each other: the data is fragmented and unstructured. Where ten years ago parents had to dig around for change for lunch money, now they have to remember the login details for umpteen different platforms. At Inspired, all the data is on one platform owned by the company.
There’s one other, crucial difference about Inspired Education’s approach. They laid a foundation, so that when LLMs hit the mainstream in 2023, Inspired had everything in place to leverage its potential for education in ways that had never been done before. And they were ready to move fast.
AI’s impact on core processes
Torben came late to the education business, after spending the first part of his career working for media companies, starting at the same time as the internet began to take off. ‘I spent 20 years on digital transformation in print and television, digitising all that,’ he recalls, ‘but that sort of finished. So I wondered, which other industry was ripe for a similar level of transformation?’ As a parent of four children, Torben could see that education fitted the bill. He wanted to be a part of changing it.
But he hadn’t foreseen the explosion of AI. Inspired’s platform had been built to use well-established adaptive learning software from third-party providers. Torben freely admits that when ChatGPT burst onto the scene in 2023, even a forward-looking company like Inspired had no generative AI strategy.
‘It hit me like a revelation,’ he says. ‘I could see it would be big in many areas of our company, but particularly the core academic processes,’ so much of which are based on the written and spoken word. ‘And the reason we’ve been able to move much faster in AI than other educational companies,’ Torben adds, ‘is because we had the platform. No-one had as much data as we did.’ Grasping the implications of GenAI, he established a series of ‘speedboat’ projects that would deliver quick-turnaround applications for Inspired’s three key stakeholder groups: students, staff and parents.
First up: the teachers.
The Lesson Planner
To understand the landscape, Faculty helped Inspired survey school leaders and frontline teachers. Over 700 responded, revealing that teachers were spending an average of over six hours per week on lesson planning, and over four hours on creating and marking weekly tests.
‘Teachers can be quite cynical about new technology,’ Torben observes, ‘and with good reasons. A lot of technology doesn’t really help them. It’s more tracking, it’s more information that somebody else needs, and it doesn’t really help that day-to-day work in terms of teaching.’ But the surveys showed that teachers were willing to embrace AI if it helped them - not only to automate and speed up repetitive tasks, but also as a way of stimulating their creativity in lesson planning, or helping them differentiate lessons for different ability levels within their classes.
The targets the Faculty team were set for the Lesson Planner project were ambitious: to build an application that could automatically generate high-quality lesson materials in a way that replicated Inspired’s teachers’ best practice, and saved them time. These lessons would have to cover every subject in the curriculum, across different topics and year-groups.
Inspired asked their teachers from around the world to share their best lesson plans, to be analysed by the AI model. Were teachers reluctant to hand over their prized materials, some of which had been refined over years of practice, to train an algorithm? ‘Not really,’ says Torben. ‘Their approach has been that the collective sharing of this information helps everybody, not just that particular teacher but also new teachers coming into the trade, who otherwise have to build everything up from scratch.’ The teachers responded with over 50,000 examples.
With those to work off, Faculty developed a proprietary LLM modelling framework for Inspired. Following best-practice learning design, it starts with the teacher. They create and edit a lesson plan with help from AI, and then type in some simple lesson objectives in a few sentences. When they’re done, the model goes to work.
Built on top of GPT4, it breaks down the content generation system into multiple steps. It uses Retrieval Augmented Generation techniques to work out which are the most relevant lessons in the databank; draws the appropriate content from them for structuring lessons; then generates the lesson content according to this plan. It creates teaching slides, in-class activities and assignments, all formatted to Inspired’s design template. Everything is ready for the teacher to review in under a minute. And because it’s all on the single study platform, lessons are available for any pupils who need to catch up, or want to review the material after class.
But as much as you try to teach them, how can you tell what’s going in?
Every time a teacher or a student engages with the lesson-planning and Cycle Test tools, the systems learn and calibrate the models, constantly improving their ability to generate appropriate lesson materials, test questions and useful feedback.
The Cycle Test Generator
Visit any Inspired school anywhere in the world at 10am on a Monday morning, and you’ll find every pupil over the age of nine sitting a ‘Cycle Test’ to assess what they’ve learned in the past week. These have been part of the routine at Inspired ever since it was founded, allowing teachers to track progress in fine detail and give formative feedback, and getting children used to the discipline of test-taking long before they reach the high-stakes exams of later years.
Taking the tests is a lot of work for the kids. It’s even more onerous for the teachers: writing the tests, marking them, entering marks in the system and providing the kind of rich feedback that makes a difference. Over 500,000 tests need to be generated, leading to two million scripts that have to be marked every year.
Enter the Cycle Test Generator. Using existing test examples, and user research with teachers, Faculty and Inspired created an AI application that covers the whole process from start to finish. It allows teachers to choose the subject and topic, then automatically generates questions in line with the British National Curriculum across the full range of question types: multiple choice, show-your-working, retrieval and (coming soon) essays.
Once the teacher has edited the test and is happy with it, they can securely share it with their students. When a student’s completed it, the proprietary marking engine marks the answers, and provides written feedback down to the question level. Everything is automatically recorded on the learning platform, so that parents and administrators can track progress. The teacher still retains the power to review, edit and configure every aspect of the process - but in a fraction of the time it used to take.
The Cycle Test Generator is being piloted in over 25 Inspired schools and is already transforming teacher workload. When it’s fully rolled out, Torben estimates it will save staff over a million hours a year, time which can be reinvested into teaching. And that’s just the beginning.
Every time a teacher or a student engages with the lesson-planning and Cycle Test tools, the systems learn and calibrate the models, constantly improving their ability to generate appropriate lesson materials, test questions and useful feedback. All of which saves teachers time, and helps kids learn.
Setting goals beyond the classroom
Torben’s goal is to cut staff lesson-planning time, homework-marking time, test-preparation and marking time all ‘in half’. In terms of pupil outcomes, Inspired has already seen an 8% rise in performance - equivalent to an entire grade boundary - thanks to its wider investments in education technology. They’re now thinking about training the AI in other languages and other curricula. And they’ve barely scratched the surface of what their data can yield.
As well as the ‘speedboat’ applications, Torben has asked Faculty to take on a ‘supertanker’. This is a long-term, strategic project to structure every piece of data held anywhere in Inspired’s systems, so it can all be aggregated and analysed for insights into how children learn, and used to correlate insights from students around the world. ‘We are probably the world leader in terms of having the amount of structured data required to do this at scale,’ Torben points out, ‘and we can use that information to supplement the learning curve for each child.’
Nor is he looking to keep the software and the insights proprietary. ‘We have these top private schools, so we would be a logical first mover investing more on this up front, but we are very conscious that this can also help the wider system. And our intention and approach is for this to escape out to the state school systems.’
It’s very likely that education will change more in the next ten years, thanks to AI, than it has in the last five hundred. As Inspired shows, it has the potential to make teachers better at what they do well, and free them from what slows them down, to the huge benefit of the kids they teach.
Everyone can learn from that.
The lesson in summary
Augment human tasks that require judgement. Automate those that don’t.
- AI will almost certainly impact most jobs. But choices about the way the technology is deployed will determine how these impacts are felt. And for many roles, there are paths you can choose which will have positive outcomes for workers who are affected.
- Most people enjoy the core of their job. It’s what they’re good at and it’s an important part of how they create value to the world. But many of these people will tell you that admin, reporting and other bureaucracy take up too much of their time, and crowd out their ability to focus on the core.
- This is going to change. The kind of tasks that AI is extremely good at map well to the tasks that are responsible for this crowding out. Well designed AI programmes build AI into workflows in a way that automates away the low-value, routine tasks.
- Thus people will be freed up to focus on the high-value tasks at the core of their roles. In most cases these tasks require professional judgement. It is usually unwise to try and replace this human judgement. Instead, AI should be used to improve the speed, quality and execution of high-value decision-making.
- This blend of automation and augmentation offers a vision for the future of work that is both more productive and more fulfilling. By the end of the decade, most cognitive workers in the economy will have many fewer low-value, routine tasks to do, and will have Intelligent Decision Support augmenting their performance of high-value tasks.
Did you enjoy this story? There are nine others just like it, told from the perspectives of nine of our other inspiring customers, in the full book 'Ten Lessons From Ten Years of Applied AI'. Just leave your details below to get instant access to your copy of the book.
With contributions from:
Ten Lessons From Ten Years of Applied AI
Download the eBook
Get instant access to ten examples of AI solving the world's biggest challenges, told through the stories of ten of our most brilliant customers.