Generation Generative: How developments in artificial intelligence are an opportunity to improve education and wider society


Economic growth is the most reliable way to improve citizens’ quality of life. This requires increases in productivity, which depend on having a well-educated workforce. However, our approach to education is not working as well as it could: the UK regularly does worse than other Western and East Asian nations in OECD PISA rankings, for instance [1]. Moreover, our system of assessment negatively effects student welfare and is being undermined by advances in artificial intelligence (AI). We can turn AI from a threat into an opportunity, by harnessing it as a tool to modernise our education and examination system and increase economic productivity and societal wellbeing. In this essay I will recommend that the Department for Education (DfE) pursue the following three proposals: (1), create a regional network of AI excellence hubs to support teachers in making the most of the technology; (2), develop a national policy on the permissibility of students using AI within coursework; and (3), explore the feasibility of using AI tools as a means of assessment. This essay suggests potential pilot implementations of these measures, as well as explaining how they would improve life in the UK for students, educators, and the wider public.


For over a decade, the UK has been stuck in a rut of weak growth, declining real terms pay, and stagnating quality of life. Though all developed nations have suffered poor productivity growth since the 2008 financial crisis, the UK has particularly underperformed. We know that human capital is a key factor of production. Yet, despite Britain’s universities being among the best globally, our education sector has not been delivering the economic dividends we would hope for. Indeed, the Times Education Commission recently concluded “the system is failing on every measure” [2]. Addressing this issue is crucial: government has a duty to ensure that today’s young people are well-prepared for tomorrow’s new era of technology – and updating our approach to education will raise living standards more widely, by enabling economic growth.

Young people can only unlock their potential if supported, academically and emotionally. When I talk with my peers at school about the obstacles holding us back, poor mental health consistently comes up as the biggest. Precarity in wellbeing is widespread – over one fifth of those aged 17 to 24 have a probable mental health disorder, a figure which has doubled since 2017 [3]. Evidence suggests that much of students’ stress and anxiety directly stems from the education system itself, with homework and exams cited as the greatest sources of worry [4]. Education is a devolved matter, but in England, worsening student wellbeing has been linked to the 2015 shift to high-stakes linear assessment: a majority of teachers believe that GCSE and A-level reforms exacerbated poor student mental health [5]. With the flaws of existing exams highlighted by the covid-19 pandemic, the time is right to find models of assessment fit for the 21st century [6, 7, 8].

It is not only students for whom the status quo is unsustainable. One of the biggest ongoing challenges for the education sector is recruitment and retention. Demanding workloads, especially in lesson planning and marking, contribute to staff dissatisfaction and harm morale [9]. When these working conditions cause trade disputes and industrial action, as seen currently within schools and universities, the disruption affects working parents and caregivers in addition to the students themselves. Even when it is “business as usual”, the churn in experienced staff – with fewer than 60% of new teachers remaining in the sector for at least ten years [10] – leads to poorer educational outcomes. Any policies which go towards easing workloads without compromising standards should therefore be enthusiastically investigated.


Some would be tempted to add recent developments in “generative AI” to the above set of problems. ChatGPT, in particular, has widened public awareness of how convincing machine-produced content can be, and the implications this has for homework or examined coursework. Although some technologies to distinguish between human and AI generation are being developed [11, 12], they are not yet robust and can be deceived by paraphrasing. However, this transformation offers an opportunity to reimagine our exam system to better fulfil its purposes. Assessment is a means of measuring pupil ability and holding the education system accountable, but more importantly, the process of study is designed to prepare young people for the working world.

That world is now changing. AI assistance is already common across a range of professions, be it software development [13], creative arts [14], or journalism [15], and will become still more routine as the technology improves further. Search engines mean that facts are already at learners’ fingertips; increasingly, explanations are too, through a growing array of sophisticated chatbots. Metacognition and an appreciation of the learning tools available will therefore become relatively more important than initial knowledge or understanding. Just as it would be remiss of our education system not to teach students how to use calculators or the Internet, so too should it now equip young people with the skills to make the most of AI in their learning and working careers.


Understandably, many teachers are themselves not yet fluent in the use of AI tools, or even aware of the technology’s existence. AI has the potential to transform the way that students learn, but this will only happen if schools are encouraged to innovate and given guidelines on best practice. Some academics are already emerging as trailblazers, using AI to address student misconceptions, facilitate learning through teaching, and help knowledge transfer between contexts [16]. AI can also enable less advantaged students to express their ideas, ultimately narrowing the attainment gap [17]. DfE should therefore expedite the diffusion of expertise around AI in education, by creating a network of school and higher education AI excellence centres that other institutions can learn from. This could be modelled on the successful, existing “subject hubs” scheme. DfE should also coordinate with Oak National Academy’s ongoing curriculum redesigns, to ensure that all teachers are supported to take full advantage of AI’s potential, and that no student misses out on its benefits.

The AI excellence hubs could also be used to trial and develop a nationwide AI coursework policy. Some institutions have already drafted guidelines on the matter: in the UK, UCL is leading the way with comprehensive advice to faculty [18], with some US colleges having similarly detailed policies [19]. However, as awareness has grown, different institutions’ approaches have diverged [20]. Although universities have independent awarding powers, a nationally harmonised AI policy would ensure consistency among equivalent qualifications. The same is true for exam board policies at secondary school level. Moreover, a national approach developed by DfE alongside Ofqual and the Office for Students would allow the UK to leverage its education sector’s reputation to lead the world in AI regulation more widely, attracting economy-boosting private investment.

Finally, DfE should work alongside regulators to explore the use of AI in assessing examined work. The 2020 grading scandal showed the importance of a trustworthy and fair exam system, so caution is of course needed to avoid potential biases. However, we must also acknowledge the existing model’s flaws: Ofqual’s own figures show that qualification accuracy is barely above 50% in some essay-based subjects [21]. This unreliability blights the lives of affected young people, potentially leaving them with incorrect grades and dented confidence for life. Using AI as an examiner’s assistant can mitigate concerns around algorithmic bias whilst helping improve fairness. A pilot programme could investigate the feasibility of different methods of training AI systems (either with full-length scripts and attached marks, or using pairs of essay excerpts alongside labels identifying which is judged to be the stronger response). The technology’s validity could be verified similarly to that of secondary school human examiners, by checking its marks match up with those given to pre-marked scripts. Once reliability is assured, the technology could then highlight the stronger and weaker parts of a given response, and partially automate the examiner’s process of deciding marks. In addition to improving fairness and transparency, making such technology available to schools would reduce staff workloads whilst raising teaching quality, by enabling teachers to better identify the components required for a model answer (often a challenge in humanities subjects with ambiguous mark schemes [23]).


A society’s approach to education determines its future. With technological development accelerating, our education system must equip young people with the skills they need in a changing socioeconomic landscape. By investing in AI within education, the UK government can improve the lives of students and teachers whilst also realising its goal of becoming a global technological leader, thereby benefitting society as a whole. Education is a comparatively low-risk area compared with other industries, and thus an ideal sandbox to explore novel applications of and regulatory approaches to AI. Embracing this technology and acknowledging the need to produce careful policies regulating its use will enable us to create a stronger education system, and a brighter future for the UK as a whole.