As John Jerrim noted in Schools Week recently, the reliability of Ofsted’s proposed new grading system is not yet guaranteed, and this is high-stakes for the inspectorate. It is crucial for building confidence in the consistency and fairness of the school review process – a challenge faced by education systems across the globe.
Our research from our work in the Middle East shows artificial intelligence (AI) holds the potential to be transformative in this regard. Our findings suggest that its judicious use can enhance the quality, validity and reliability of inspections.
We have been testing the feasibility of AI in school reviews through a series of over 150 laboratory simulations and a small-scale field test, using the United Arab Emirates’ inspection framework.
Our trials have shown AI can accelerate many of the processes, enhance the breadth, depth and triangulation of evidence and potentially help guide novice inspectors through inspections.
We believe that an approach which allows more time for experts to evaluate to the best of their abilities promises greater reliability and validity of judgements.
The sheer volume of information inspectors must process in a limited time is significant. Indeed, one of the concerns that has been raised about Ofsted’s report cards is that inspectors will need to assess a larger set of performance categories within the same time period.
From our trials, we can see that the time savings AI can provide are substantial, reducing tasks which can take hours to seconds.
For example, synthesising large amounts of messy data and providing summaries for different audiences during the inspection process are currently time-consuming tasks which can with care be largely automated.
Every hour taken away from these tasks during inspection is an hour which can be spent refining the team’s understanding of each provider’s context.
There is a strong case for implementing the technology now
The model we developed also helps analyse the quality of the evidence base during the inspection, triangulating the emerging data against the relevant framework and accurately identifying how the evidence stacks up against a judgment. It flags areas that warrant additional attention from the inspectors, ensuring a more thorough and therefore more accurate evaluation.
As with AI in search engine results, it is possible to quickly drill down from the AI’s conclusions to the evidence it has drawn on. We argue that this improves the traceability of evidence in inspections, in turn helping to increase transparency and trust in the judgment and the inspection system itself.
We have also begun to explore how this technology can be adapted to support inspection teams in real-time. An AI assistant-type application might help steer inexperienced inspectors through the procedural steps of inspection, as well as directing attention to specific areas of the evidence base.
This allows different choices in resourcing. Inspectors who are not specialists in a particular subject or phase can access real-time prompts and contextual information courtesy of AI. This means all areas of a school’s performance can be evaluated rigorously and fairly, regardless of the individual expertise of the inspection team.
Equally, it might allow domain specialists with limited inspection experience to examine aspects such as inclusion and wellbeing in depth without risking the overall inspection quality.
We have developed a clear way forward for AI in school reviews and are excited by its enormous potential to enhance quality, efficiency and fairness.
The promising results from these experimental trials suggest that while further research is needed to understand how AI-driven inspections perform in different school contexts and under varied accountability frameworks, there is a strong case for implementing the technology now.
While nascent, we strongly believe that there is a huge amount of potential for AI to enhance the quality and impact of school inspections.
This isn’t about AI replacing human judgment. It is about creating the time and opportunity for inspectors to engage with school teams and to really understand the nuance of each context.
There is therefore an opportunity for Ofsted (and school inspectors across the world) to consider the opportunities to continue to enhance and improve inspection experiences.
Your thoughts