As a school staff member, data protection officer, data protection lawyer and AI governance practitioner in the education sector for more than eight years, I have observed both the transformative potential and challenges of implementing technology in education.
The ongoing debates surrounding educational methodologies – such as the reliance on exams to benchmark student performance, and the balance between STEM subjects and the arts – have only been made more complex and urgent when considering the challenges of integrating AI responsibly into traditional educational frameworks.
However, in the realm of AI and edtech, there’s a notable gap between the focus areas of AI governance professionals and educators.
While AI governance and risk-based professionals often concentrate on technical harms, such as privacy violations and algorithmic bias, educators are primarily concerned with the broader impacts of AI on student learning and social development.
To address these diverse concerns effectively, we must centre both governance and educational perspectives in the design and development of AI systems used in schools.
By involving educators in the conversation about AI risks, technology developers can gain insights into the practical and pedagogical (the theory and practice of learning) implications of AI tools, ensuring these technologies support educational goals without compromising on technical safety and efficacy.
Similarly, tools developed by passionate educators can be built in a way that ensures harms are minimised from the outset – rather than trying to retrofit privacy, intellectual property, security or other fixes, which can prove costly or even impossible.
The current landscape of edtech uptake
To borrow a phrase from a well-travelled TV advert, AI presents a world of opportunities for education. This is reflected in the Department for Education’s statement that “AI presents exciting opportunities to improve people’s lives … AI can ensure that every child and young person, regardless of their background, is able to achieve at school or college and develop the knowledge and skills they need for life.”
In Browne Jacobson’s latest School Leaders Survey – which captures the views of leaders representing about 1,650 schools that are collectively responsible for nearly one million pupils across England – half of respondents said they’re using AI tools “regularly” or “often”, with one in five leaders implementing personalised learning in some form.
Personalised learning platforms include Duolingo for Schools, which uses sophisticated algorithms to customise language learning experiences, offering practice sessions that are tailored to the learner’s proficiency level and personal goals. However, this service is simply a management layer sitting above the user app, not a specialised product designed for use in an educational setting.
The uptake of edtech can be significantly influenced by vendor-led demand. Often, school staff are unable to articulate the specific needs for particular products, driven instead by a notion that “every other school is doing it”. This trend has led to the adoption of hastily assembled products that may not adequately meet educational needs or uphold necessary standards, particularly in data protection and security.
As we now face similar trends with the integration of AI in education, it’s crucial to learn from past experiences with edtech. The allure of adopting new technologies shouldn’t overshadow the importance of rigorous evaluation of both legal compliance and pedagogy to ensure these tools genuinely enhance educational outcomes and comply with legal requirements.
Personalised learning shouldn’t remove the human
As an AI governance professional, I was delighted to attend the Human Intelligence in the Age of AI conference, hosted by King’s College London in early March. This event provided valuable insights from experts like Sir Anthony Seldon, Head of Epsom College, and Professor Rose Luckin, Founder and CEO of Educate Ventures Research, a company that supports the education sector in leveraging AI ethically and effectively.
The speakers explained that highly personalised learning is often marketed as a “frictionless” process, but there was a real risk to pedagogy, as educators could invertedly remove a key ingredient in what makes learning truly effective and rewarding.
They emphasised the need for AI to amplify human intelligence attributes and support deeper cognitive processes. Reflecting on the conference discussions, it’s clear that while AI can make learning processes more efficient, but it’s essential to preserve the intellectual struggle that is crucial for deep learning.
As a lawyer, most of what I learned about trigonometry at school has been forgotten, but just because I didn’t become an astronomer or architect, it doesn’t mean it was pointless.
The concept of “productive friction” – the mental and emotional challenge involved in learning – parallels my experiences in the legal and AI governance fields, where deep engagement and critical thinking are indispensable.
The broad curriculum contributes to our human development and makes us more rounded individuals. Those focused on traditional AI risk could neglect to draw out these risks, instead focusing purely on technical risks of AI use.
Lessons for schools
These learnings don’t just apply to the classroom. The findings of our School Leaders Survey indicate the most common applications of AI revolve around driving efficiencies for teachers and administrative staff.
This is similar to many commercial organisations, which are implementing AI for processing large amounts of data, creating generic or repetitive content such as emails and newsletters, and interacting with customers.
But it’s important to take a step back at this point and consider expectations of the intended audience.
Communication is a fundamental instrument in building relationships between humans. Interfering with this may do more harm than good.
Parents will quickly see through AI-generated emails and could question why they aren’t worth a personally-curated message, while AI-driven feedback may not provide students with the same level of motivation as a handwritten note at the bottom of a submitted piece of work. Again, this is something that traditional risk assessment may overlook.
Collaborative innovation is path forward
It’s understandable that schools won’t want to feel like they’re being left behind and will want to actively seek out the best resources to enhance their pupils’ learning experience.
But they shouldn’t be swayed by pressure from other schools taking a lead in technology adoption or from vendors pushing forward exciting new products. AI governance professionals should also listen carefully to educators and be open to the exciting potential for AI use, rather than viewing projects entirely through a risk-based lens.
Both points of view are equally valid in the debate around the future of AI in education and it has never been more important to ensure that there is collaborative working in this space.
Claire Archibald
Claire Archibald is a legal director at UK and Ireland law firm Browne Jacobson. She specialises in advising education clients with artificial intelligence, data protection, information governance and freedom of information matters. She is an experienced data protection professional, having led a data protection officer (DPO) service for schools and academies in a local authority setting, providing external DPO services to about 400 settings across England.
