2024 is expected to usher in a more down-to-earth and tangible side of generative AI, following the hype of 2023. With the rapid deployment of AI across all sectors, it is more important than ever to think critically about where and how the technology can best be used.
In an interview at the recent ISE conference in Barcelona, Mhairi Atkins, Ethics Fellow at the Alan Turing Institute, highlighted the need to ’make sure that we are using technology to address existing problems’, rather than just integrating it for its own sake.
In education, one of the most stubborn existing problems is the attainment gap. This refers to the statistical difference in academic achievement between students from more privileged and less privileged socioeconomic backgrounds.
According to a Nuffield Foundation funded IFS study, published in 2022 using data collected from UK schools, less than 50% of students from low-income backgrounds (measured by whether a student qualified for free school meals) had reached expected levels of achievement at the end of primary school, versus 70% of students from better-off families.
Secondary schools showed a similar trend: just 40% of disadvantaged students attained good GCSEs in Maths and English compared to 60% of their peers.
This pattern is not unique to the UK, but has been found across education systems globally to varying extents. Yet despite the global scale of this problem, and many attempts to tackle it through a range of policies and community initiatives, it has nevertheless remained a stubborn trend that propagates a deep-seated social divide between the rich and poor.
Since the Covid pandemic, when students were forced to study remotely at home, the attainment gap has widened. This highlights the impact of a student’s home life on their academic achievement, and indicates that while digital technology has opened up a more flexible and accessible form of learning, it has not yet managed to level the playing field in education for disadvantaged students.
Looking to the future, which is likely to bring increasing forms of digital learning and flexible/remote study options, the attainment gap may keep on widening unless measures are taken to equalise digital learning opportunities, both at home and at school.
Current initiatives for educational equity
Access to a personalised learning experience is one of the key privileges that gives students from wealthier backgrounds an academic advantage over their peers. For example, students whose parents can afford private school are given more individual attention and support from their teachers due to smaller class sizes. Private one-to-one tutoring is another popular way to boost a child’s academic achievement that has traditionally only been affordable for wealthier families.
In the UK, there are already measures that attempt to make personalised learning available to students from less fortunate backgrounds. For example, many private schools offer some level of funding through scholarships and bursaries for talented students whose parents are not able to afford the full tuition fees. It is also common for state schools and community learning centres to offer after-school clubs which aim to provide disadvantaged students with more individually targeted support and learning opportunities.
However, these initiatives are often underfunded and ineffective at closing the attainment gap. Scholarship programs, for example, typically function more as a talent scouting mechanism than anything else. Meanwhile, after-school clubs are not usually a sufficient substitute for the better quality support that is provided by private tutoring. This is particularly the case in highly deprived areas, where greater demand for support puts a heavy strain on resources which leads to a deterioration in their quality.
Despite these challenges, evidence suggests that making personalised learning accessible to students from lower socioeconomic backgrounds is one of the most effective ways to close the attainment gap. In 2020, the government launched the National Tutoring Programme (NTP) to help counteract the detrimental impact of the Covid-19 lockdown on education. This provided subsidised one-to-one tutoring for children whose families would not have been able to afford private tutoring, with the government covering 60% of the cost, and schools covering the rest.
According to research published by the Sutton Trust last year, this has helped to close the gap between higher and lower income households in terms of access to tutoring. While higher income households were 15% more likely to have access to private tutoring than lower income ones, this gap closed to just 1% when it included tutoring funded by the NTP.
The benefit of increasing access to one-to-one or small group tuition has been widely recognised, with 79% of school staff reporting that the NTP met pupil’s learning needs while still aligning with the school curriculum, according to a review published by the National Foundation for Educational Research. Senior leaders, teachers, and tutors have also confirmed the positive impact the programme is having on student attainment, progress, and confidence.
Nevertheless, the provision of tutoring for all students in need of extra support comes at a high price, and this is why the government is currently unable to commit to the NTP as a permanent scheme. 2024 is scheduled to be its fourth and final year, after which tutoring will once again become private and out of reach for most socioeconomically disadvantaged students.
There is hope, however, that the government may continue to subsidise some form of a public tutoring scheme, given the success of the NTP over the last four years. This comes after the government agreed to increase funding from a planned 25% coverage of the costs this year to 50%. This is “clear evidence of [their] commitment and a down payment on long-term economic growth”, according to John Glen, Chief Secretary of the Treasury at the Tribal group, a collaborative partner in the scheme. There is also widespread support for the scheme’s continuance from educational advocacy unions such as the Sutton Trust and NASUWT.
The long term sustainability of schemes such as the NTP hang on their cost effectiveness. If the NTP is renewed for future years, it will likely be with a significantly reduced budget for funding due to constraints on public spending. This is where AI can step in. If deployed safely and thoughtfully, AI could prove instrumental to the future of education by reducing costs for such schemes, thus equalizing access to personalized learning on a permanent and financially sustainable basis.
AI tutoring and its cost-cutting appeal
One of the greatest appeals of using AI for publicly funded tutoring programs is undoubtedly the extent to which it can cuts costs. Khanmigo, one of the top digital tutoring platforms, offers access to its service for just $4 per month per student as a non-for-profit organisation, which contrasts to private tutoring rates of around $30 per student for just one hour. This minimal fee could easily be covered by the government, schools, or even parents who are on a budget. On a broader scale, this could save governments and educational institutions thousands while still providing students with personalised learning support in the form of digital tutoring.
While AI tutor platforms cannot of course replace the unique value of human tutors, they can nevertheless still provide a relatively comprehensive and personalised learning experience. For example, Khanmigo and Cognii, another digital tutoring platform, provide a step-by-step explanation of school subject topics, give students individualised feedback on their answers to questions, and can support literacy through a set of dialogue-based activities such as word games, debates, and the co-writing of stories.
According to Paul Henninger, head of UK Connected Technology and Global Lighthouse, AI tutoring platforms such as these will be particularly transformative for low achievers, by providing them with readily available feedback whenever they need:
“AI can be used to provide high quality automated feedback on exercises, for example where a student would otherwise have to rely on the limited time available from a real tutor or teacher. In the longer term that’s likely the best way to get to better learning outcomes for low achievers.”
Paul Henninger, head of Connected Technology and Global Lighthouse at KPMG UK
Other AI tutoring applications such as Duolingo and MATHai are providing free or low-cost learning support for specific academic skills: Duolingo for language learning, and MATHai for numeracy. These applications, which specialise in particular subjects, can be particularly useful for disadvantaged students who are struggling with a particular subject, but whose overall academic achievement is not low enough to qualify them for subsidised tutoring.
They also open up opportunities for disadvantaged students to expand the breadth of their education by getting a free introduction to subjects that are not taught in school, such as minority languages. Typically, access to non-mainstream subjects has been a privilege for students from wealthier families who can afford to pay for subject-specific private tutoring or extra-curricular classes/activities.
Overall, AI-powered tutoring support, while still limited in comparison to human tutoring, can reduce the workload of tutors through the automation and gamification of simpler and lower level learning activities. This may change the role of human tutoring to focus more on soft skills such as critical thinking, creativity, and problem-solving, which could mean that one-to-one tutoring sessions with students is needed less frequently. The integration of AI tutoring platforms could thus be a key way for schools to reduce the overall cost of tutoring per student, and make publicly funded schemes such as the NTP more financially sustainable.
Future considerations (and how to avoid more digital division)
Despite the expectation for AI to cut tutoring costs in a relatively clear-cut way as described above, it remains difficult to predict the wider impact of AI in educational equity. For example, as the use of AI tutoring platforms becomes more widespread, access to one-to-one human tutoring may become more sought after and needed than ever, with increasing recognition of the limitations of virtual tutors in comparison to human ones. This is true not only in terms of academic expertise, but also in terms of improving students’ holistic wellbeing by connecting with human tutors.
Indeed, sociological research suggests that one-to-one/small group tutoring is effective not just because of its academic support, but also because of the emotional and mental health support it provides for students, helping them to manage anxiety, boost their self-esteem, and increase motivation. This latter type of support has been found to be particularly beneficial for disadvantaged students who may be more prone to mental health issues due to instability in their home lives, or who may lack academic confidence due to their social class, as discussed in a study by PhD researcher Annette Lareau.
Michael Trucano explores the possibility of how AI could affect educational equity (or not), in an article published by Brookings, a public policy think tank in America. He suggests that AI could create another ‘digital divide’, where advantaged pupils would have access both to the technology and the people who can help them use it, whereas disadvantaged pupils would have access only to the technology itself.
This reflects an existing trend in AI’s development where as the technology has become more accessible and widely used, the skills and expertise needed to deploy the technology effectively, such as coding, programming, and data management, have become increasingly sought after and expensive. In education, this would mean that as personalised AI tutoring became a new basic standard of education that was accessible to all, private human-directed tutoring would once again become a way for wealthier families to give their children an academic advantage over their peers.
Could this be avoided?
Ensuring equity in access to educational resources and support is likely to be an ongoing battle as AI continues to provide more educational resources and digital forms of learning support. However, with extensive deployment of AI technology expected for the year ahead, now could be the time when an equitable and forward-thinking approach to integrating AI could make a real difference to the success of digitizing educational systems over the next few years.
Measures such as ensuring that AI tutoring platforms are integrated alongside human tutors, and ensuring that all students are taught about the most effective ways to use AI, could help ensure that the technology is harnessed for the greater good when it comes to closing the attainment gap in education. Educational equity schemes such as the NTP provide great pilot opportunities for AI to be integrated into education in responsible and censored ways that promote the ethical use of technology in helping to create a more equal society.
