The expectations to use AI seems to be growing. Business adverts, social media and news articles herald its ability to reduce workload, unconsciously suggesting that we should all be using it. The promise is that it will automate tasks, improve student feedback, and provide valuable student insights. Whilst it may save time, that time will no doubt be filled with something else. But this aside, integration of AI into education could possibly increase workload. Here are a few ways of how this might occur:
Content: Whilst AI can quickly curate content and produce great resources, teachers need to spend additional time selecting and vetting. Even with the most carefully considered prompt, smooth sounding AI generated content does not always align to a given curriculum or is marred by inaccuracies. When I’ve prompted for challenging content that is specific to GCSE Computer Science for a particular exam board, there has been considerable irrelevant output. Think of it as like mining for gold. AI can help uncover golden nuggets, but it can take a lot of work to separate them from the ore. Whilst the overall quality of resources is likely to increase, the resources may take longer to develop. This issue helps to strengthen the idea of centralised resource generation – where resources are generated by a central body that then shares them between all schools, thereby providing a level playing field and reducing teacher workload.
Assessment: This is where I think the allure of time-saving nature of AI will increase workload the most. AI-powered systems can create considerable feedback on student work, much of which is incredibly useful. However, teachers will need to check both the student work as well as the AI feedback. Where the AI-feedback has fallen short, teachers will then feel it necessary to write feedback of a similar length to that generated by AI, which by and large is more than what a teacher would normally be expected to write. Demand for increasing student feedback could be driven by a few factors: teachers wanting to give their students the best personalised feedback, students who get used to receiving an increased amount of feedback, and parents who may start to demand use of technology to support their child.
And by the way: “Education institutions must not allow or cause pupils’ original work to be used to train generative AI models unless they have appropriate consent or exemption to copyright. Consent would need to be from the student if over 18, and from their parent or legal guardian if under 18.” (DfE Generative AI in Education, October 2023). By using LLMs to generate feedback on student work, teachers can be unwittingly providing it with data which is used to train and refine the model which is goes against DfE guidelines.
Training: There are now increased training sessions based around teacher and student use of AI. This is on top of existing training sessions. Teachers need to be kept lockstep with evolving AI, having to learn about what they should and should not do with AI (especially in relation to data protection), how to guide effective use of AI by students, and what action to take when AI is used inappropriately. This additional responsibility adds to workload, and more so if staff training is inadequate.
And by the way: Ofsted has very clear principles about expectations surrounding the use of AI. I recommend looking at table 2 in Ofsted’s approach to AI (April 2024).
Data Analysis and Interpretation: There is no doubt that AI can generate vast amounts of data related to students’ learning progress, performance, and behaviour. Complex data takes time to analyse and translate into actionable strategies.
In summary, while AI holds great promise for improving education, its implementation may increase the workload of teachers. To mitigate these challenges, it is essential to provide teachers with adequate training.
