红桃影视

To a broadly positive reception, Gillian Keegan to announce her clearest position to date on the role of Artificial Intelligence (AI) in education. Had she done so in a room full of teachers or school leaders, perhaps the reaction would have been slightly more sceptical. Because as much as she rightly claimed AI could 鈥渞adically reduce the amount of time teachers spend marking鈥� and 鈥渢ake much of the heavy lifting out of compiling lesson plans鈥�, I suspect a number of Schools Week readers would have been sat there asking, 鈥淗aven鈥檛 we heard this all before?鈥�

Teacher workload has become one of those perennial issues every secretary of state for education tries to tackle early on in their tenure. That鈥檚 because they know it鈥檚 having a devastating impact on recruitment and retention, and because they鈥檙e regularly presented with analysis and evidence from DfE civil servants on the drivers of the problem. Whichever party wins the next election, it鈥檚 not an issue that鈥檚 going away any time soon.

So it isn鈥檛 that they don鈥檛 鈥榞et it鈥� or that they don鈥檛 have a good enough grasp of the issue. Rather it鈥檚 that the solutions chosen to date have not yet been meaty enough to match the scale of the challenge. For example, new and tweaks to the accountability system, while well-intentioned, have not been enough to make a difference.

And so many of you have probably been left thinking, 鈥楬ere we go again, this is just more of the same.鈥� But this is where I disagree.

While there are risks emerging from AI, particularly on the ethics of how it鈥檚 implemented, there are dramatic opportunities too.

In particular, Keegan is absolutely right to focus on marking. As highlighted in their 2016 report, marking was 鈥渢he single biggest contributor to unsustainable workload in the Department for Education鈥檚 2014 Workload Challenge 鈥� a consultation which gathered more than 44,000 responses from teachers, support staff and others.鈥�

The solution? AI could be used to automate simple marking processes and provide better or quicker feedback and information. That 诲辞别蝉苍鈥檛 mean AI reading lengthy essays and providing complex feedback all on its own; We know that this is best done by humans and that many flaws in current AI systems could risk inconsistency or even prejudice.

But where we have 鈥榗losed鈥� questions (for example, defining scientific terms, factual questions about an historical event or simple arithmetic) AI absolutely has the potential to speed this marking process up. On feedback, this 诲辞别蝉苍鈥檛 mean replacing the judgement of a teacher or pushing a small selection of standard comments back to the student for every piece of work they do. Rather, it means helping teachers with shorter pieces of feedback, and supporting them to analyse and distribute information quickly to senior leaders and parents.

AI should be seen as aiding teachers rather than replacing what they do. So rather than marking a lengthy essay response, it could instead highlight relevant information related to the mark scheme. And where used properly for summative exams, AI could actually reduce bias and increase fairness, by providing an additional layer of checks on individual markers and moderators.

As well as marking assessments, AI can also be used to formulate them. For example, a bespoke on-screen quiz on a topic could be created, and the AI could then mark any responses and provide a feedback commentary on these. Not only would this help ease teacher workload, but we believe it could also help actually enhance the student learning experience and environment. 

Of course, many of these benefits are still hypothetical, and some scepticism will quite rightly remain until teachers and leaders see hard evidence of the benefits. At AQA, we鈥檙e determined to do just that 鈥� to work with the sector to crack the teacher workload problem once and for all.