It's not just about the teaching: how generative AI will transform school operations
A lot of attention has been directed at AI's creative chops. However, as a tool to sort large amounts of messy data, it's super powerful. Let's look at five potential uses for the back office.
Most of the educational world is looking in one direction when it comes to the power of generative AI. As a tool for generating ideas, planning and resourcing it’s already adding huge value. As a creative tool it has great promise. And it won’t be long before every student has their own AI that learns and grows with them, being a supportive companion in their academic and social development.
But these miss one of the most powerful use cases for AI, and that is as a tool to organise and analyse messy data. And there’s plenty of that in schools.
Generative AI is a rule follower. It has no free will, no ability to think for itself. Its judgment doesn’t get clouded, it’s not prone to emotional outbursts (Bing Chat’s initial weirdness aside), and it will never question authority. It is, in many ways, the ideal back office administrator, machinelike and consistent.
It’s usually the schools where students make the most progress that are the best organised in terms of operations. Academic leaders need for things to happen in the background: salaries to be paid on time, the right teaching supplies to be bought before term starts, IT systems that just work, and so on. It’s when any of these fail so well that teacher morale can be eroded and trust between the academic and operational teams can break down.
There are a number of ways in which generative AI can help school operations right now, and as it develops it will only become further ingrained into the daily running of schools. In time we’ll see AI sitting in the background doing many menial administrative tasks, a good number of them automated. Let’s look at a few examples of how it can be used both now and in the near future.
Reporting
Schools generate a lot of data, and periodically this needs to be sorted and analysed. Whether end of term school reports, reports to the board, or weekly attendance summaries, it’s important that the right people see evidence that the school is heading in the right direction, and what governors, leaders and teachers are doing about it if that’s not the case.
Writing and checking progress reports at the end of each term takes up a lot of teacher and leader time. To make every report bespoke to each student can be a struggle, particularly if you teach subjects like music and art where you may see almost every student in the school once or twice a week. Teachers therefore tend to fall back on stock phrases which can sound wooden and insincere. The best reports are personalised and show a connection between the teacher and student. However, all too often rushed teachers resort to copy-paste, which can be embarrassing when pronouns are incorrectly used, or (even worse) the wrong name is shown. We’ve all been there.
But if we think about exactly what reports are, they are a summary of what the student has done that term and the progress he or she has made. They were there at the start, they’ve done this, and now they’re here. And if an AI assistant has access to all that information, both through the assessment of student work and student grades, it will not be too difficult for it to write a far more detailed and accurate report on academic progress, leaving the teacher to write a comment about how the student has been in class, any welfare concerns and so on. Or, to leave those comments for a face to face meeting, which is a far more effective way to communicate with parents. Two way is always better than one way.
We can take this a stage further, and question whether we need periodic progress reports at all if parents have access to a portal where they can generate their own reports whenever they want. I am currently working with a development team on a prompt-driven management information system which will allow for any user to input a prompt and to have the AI generate a response. You can imagine the parent simply having this sort of conversation with the AI system:
Parent: Can you tell me how James has been getting on in maths this week? I know he was struggling with the topic.
AI: He’s been doing better actually. He scored 80% on his last test which is 20% up on the week before. He also seems happier: when I was chatting with him earlier he mentioned he was getting more sleep. I think you being firmer with him on using his phone at night has helped!
Parent: That’s great to hear!
An AI system that not only has access to data but also user interactions (both student and parent) can generate rich analysis, and the natural language capabilities of generative AI can wrap this analysis into a friendly conversational tone that can reassure parents that the school has their child’s education well in hand.
Conversely, if progress isn’t being made, or it is clear a student is struggling with a particular subject (or indeed teacher), then the AI can politely suggest the parent go into the school for a meeting. This can work in the other direction, with the school being alerted by the AI if there is a sudden fall off in progress across subjects, as this might suggest problems at home that may need to be explored.
The often siloed nature of subject teaching means that students can sometimes fall between the cracks if there isn’t sufficiently robust internal monitoring across classes. Sadly it’s only when something more serious happens that some schools suddenly wake up and take notice. Thankfully it’s less of an issue nowadays, but there are still examples where poor communication can lead to catastrophic results.
As I touched on in a previous article on the changing role of the school leader, AI will soon be generating a lot of the daily reports leaders need in order to make effective strategic decisions. Whether these are analyses of teacher effectiveness, student attainment, or progress against development plan objectives, the AI will be able to do it all in the background, creating easy to read reports that can be used to inform leadership actions and as evidence of progress against standards to the board of governors. As we have already seen, this may change the nature of school leadership, removing much of the data management and administration and focusing more on teacher experts, project-based learning architects, and project managers.
The key here is that AI is great at making connections. It can take large amounts of data from various sources and look for patterns. And whilst we are mainly seeing that being used to support things like planning and writing right now, once the ChatGPT-4 Code Interpreter plugin is released (currently in Alpha testing mode with a small number of developers) we will see a lot more use cases for generative AI as a pattern analyser. Once we have that, I think we will be in for a surprise as to how many rich insights we will get about the factors that impact student progress and be able to tailor solutions in response.
Admissions
The admissions department in any school is a critical function. This is particularly the case in a private school, where the first interface with the school can determine whether a fee-paying parent wishes to continue with the process or not. As one parent told me a few years back, in a school I had recently taken on as interim Head: “If you can’t get the basics right and don’t even bother to contact me after I registered on your website, how can you expect me to trust that you’re well run enough to educate my children?” I was able to convince them that we were well run enough, but I shouldn’t have had to. If the admissions team had done their job properly, following up on the initial website enquiry in good time, that wouldn’t have happened.
It’s not only fee-paying schools where admissions has to be tight. Any transition to a new school is a stressful experience for both students and their parents, so ensuring families have the right information at the right time is vital. From the moment the process is kicked off to the moment the child is sitting in the classroom, everything should run like clockwork. Sadly this is not always the case, particularly in smaller schools where the admissions officer might also be the front office person, the marketing lead, and the IT manager.
AI will play a large role in the future of managing admissions. There are already plenty of tools out there that can automate the process from first enquiry through to visit. Applications such as Zapier enable you to create automated ‘Zaps’ that are triggered when a user interacts with the system, for example sending a welcome email when a user submits an enquiry on the website. Very few schools I have worked in use client relationship management (CRM) tools, but I believe they will become much more common as AI takes away much of the complexity of using them.
Presently, tools like Zapier confront you with a bewildering array of options. To any marketer it is probably second nature, but for time-poor school administrators they are likely to balk at the requirement to spend hours learning how to create multi-step automations. However, AI is already finding its way into these tools, and you can now prompt Zapier the process you want and it will attempt to create the steps for you. It doesn’t always work, but as AIs become more advanced it won’t be long before we can simply write a prompt such as: “Create me a workflow that sends a personalised welcome message to a prospective family, taking all the detail they input into the enquiry form and crafting the email using the school’s house style.” Once done it will sit in the background and send out emails without any further input.
You want to reach the point where the only thing your admissions officer is doing is having phone calls and visits with prospective families. Everything else, from sending emails to organising calls and visits, should be done automatically in the background. The use of AI, to ensure the tone of emails is friendly and warm, is vital, as parents need to feel that they are being cared for from the start.
There are other ways in which AI can support the admissions process. The use of micro sites is becoming popular in private schools. Companies such as Kampus 24 produce bespoke micro sites to sell the school to prospective parents: these sites tailor images, text and video to the parent, for example focusing on the school’s sporting provision if the child is interested in sports, or its orchestra if the child is musical. They can also come with a personalised welcome from the Headteacher for a truly individualised experience. We can imagine how AI can add tremendous value here: as soon as the parent inputs information into the enquiry form and hits send, the AI can auto-generate a micro site for them, which could even be narrated by the Head, using voice cloning. It could be as if the Head is speaking to the family, referencing the child’s name as he or she talks about what the school can offer.
Through both automating background functions and personalising the admissions process, AI will add tremendous value and enable the human element of the process to focus on building relationships, which is the most important part of the admissions process.
Timetabling
Any seasoned timetabler will tell you that the process of scheduling subjects, rooms, teachers and students is as much art as it is science. I know of timetablers who use computer programmes but also those who prefer pins on a board or even post it notes as organisational tools. Again, the nature of AI as an organiser will be invaluable when it comes to optimising timetables.
Being able to take every student’s lesson allocations, every teacher’s subject specialism and non-contact time, the shape of the day and rooming options, the AI will be able to come up with the best combination of them all, and will be infinitely ‘tweak-able’, enabling schools to focus on learning optimisation for certain age groups (for example having core lessons earlier in the day for younger students or slightly later in the day for older students whose brains take longer to get going in the morning). Being able to input variables on the fly and have the timetable be fully rewritten will be possible, as will varying when students have lessons week on week, so that some subjects are not prioritised over others (for example, the French teacher complaining that her year 9 French lesson on a Friday lesson 5 is an accident waiting to happen).
But more than this, from a commercial point of view timetabling software is nothing more than an enterprise resource planning (ERP) mechanism. Its purpose is to most cost-effectively allocate human resources across rooms and across the week. And because people are the most expensive component of a school’s cost base, it makes sense to optimise their use. Human timetablers sometimes struggle with this, both due to not seeing options available, but also the subjective nature of human intervention in any process. Sometimes, one more vociferous teacher’s requests can skew an entire timetable; the classic case of the tale wagging the dog. Removing humans from the first part of the equation can remove any accusations of lobbying for the best room or the best times in the week to teach. That said, if there are variables which cannot be ignored (such as a teacher who needs take their child for medical treatment once per week at a certain time) then these can be added to the algorithm to make the necessary changes.
Procurement and Resource Allocation
The Internet of Things (IoT) and Radio Frequency Identification (RFID) tags, combined with AI, could play an important role in the efficient use of school resourcing. RFID tags, attached to school resources such as textbooks, equipment or supplies, can be easily monitored by an AI-powered ERP, enabling for monitoring of use and re-ordering when required. Resources can be tracked in real time, which could be a boon for schools that often find things going ‘wandering’. We can do this without AI, of course, but tying all of this together into one intelligently controlled system takes away the need for human intervention.
IoT devices and RFID tags continuously collect data about resource use. This can include information like how often the resources are used, by whom, and at what times of the day or week. AI can process and analyse this data, spotting trends and providing valuable insights into the most efficient use of school assets. Historical data, combined with real-time information, can help the system predict future resource needs and make recommendations.
By tying all data points together, an AI system will be able to predict future human and physical resource needs based on enrolment forecasts. This will make budgeting a much simpler process, as once the AI has enough granular data on historical consumption (per student use of photocopy paper, pens per teacher, student teacher ratio and so on) it will be able to accurately predict requirements for the following year based on expected enrolment. If this is tied to an automated resource ordering mechanism then we can move away from bulk ordering in advance towards ordering nearer the time, thus only ordering resources required.
Finally, IoT sensors integrated with RFID tags can monitor the condition and performance of school assets (like photocopiers and laptops) in real time. AI can analyse this data to identify patterns and predict potential issues before they occur. By proactively maintaining physical assets, schools will avoid the issue of broken laptops impacting learning, or teacher fury at the photocopier once again breaking down.
Financial Management
AI is already impacting on the financial world more broadly, and it won’t be along before we see AI-powered ERPs effectively supporting school budgeting and accounting practices.
AI algorithms can analyse historical financial data, including revenue, expenses, and funding sources, to identify patterns, trends, and correlations. This analysis can provide insights into the financial health of the school, revenue streams, expenditure patterns, and areas of potential cost savings. This can support both the day to day running of the school but also forward planning, as historical data can allow for scenario modelling. We have been able to run scenarios through Excel for many years, but by giving this over to AI we can see the financial impact of varying an almost infinite number of assumptions.
As we have seen through the importance of optimising the school timetable, by considering such aspects as funding restrictions, staffing needs, operational expenses and capital investments, AI can generate an optimised budget and 3-5 year business plan. This will enable school leaders to make efficient use of resources and align budgets with the school development plan. By running AI-supported risk analyses on this plan, leaders can see the financial impact of their decisions and amend planning to reduce risk where possible.
AI will also be able help with monthly management reporting, generating real-time reports and alerts to track budget variances, highlight areas of concerns, and provide insights to support financial decision-making. This will enable leaders to make timely corrections and ensures accountability to governors.
These are just a few of the ways we will begin to see AI impacting on the daily operation of schools. Data is often poorly managed leading to expensive inefficiencies, so through optimising both human and physical resourcing, managing the budget, improving admissions procedures and more effectively reporting on all of these, schools should see significant increases in efficiency and related cost savings.