AI in Education
Artificial intelligence is changing education at an extraordinary pace. Students can now access personalised explanations, receive instant feedback, and generate revision materials tailored to their individual needs, all within seconds. As these systems become increasingly sophisticated, we're beginning to see more ambitious proposals for AI-led education, including classrooms and tutoring platforms where artificial intelligence takes on much of the responsibility traditionally held by teachers.
The appeal is understandable. One of education's greatest challenges has always been providing individualised support at scale, particularly in classrooms where a teacher might be responsible for thirty students with very different abilities, needs, and levels of confidence. AI offers an exciting opportunity to address this, making personalised learning significantly more accessible and affordable.
However, I think there's an important assumption underlying much of the enthusiasm surrounding AI-led education: that the effectiveness of teaching is largely determined by how well we can explain information, and the degree to which AI is able to connect with the learner.
Anyone who has spent time teaching will appreciate that learning is considerably more complicated. A student's progress is influenced by their confidence, motivation, relationships, and previous experiences, often in ways that are difficult to recognise through their academic performance alone.
As we explore what AI could mean for the future of education, I think it's worth considering some of these less obvious aspects of learning, and what we might risk overlooking in the process.
Understanding a student goes beyond understanding their answers
Imagine a teacher explaining a new mathematical concept to a classroom of thirty students. Towards the end of the lesson, they ask whether everyone understands, and most students nod or offer a reassuring yes.
An experienced teacher, however, might interpret the situation rather differently. Perhaps a normally enthusiastic student has been unusually quiet, while another has spent the lesson copying examples without attempting any questions independently. A third might have answered everything correctly, yet appears uncomfortable when asked to explain their reasoning.
Each of these students might perform similarly on a short assessment, but their individual needs could be completely different.
Over time, teachers develop an understanding of these differences through countless small interactions. They notice changes in behaviour, recognise patterns in how students respond to challenges, and gradually learn to interpret the subtle signals that students communicate, often without consciously intending to.
And this is where I think AI-led education faces an interesting challenge.
Modern AI systems can analyse responses, identify recurring mistakes, and remember previous interactions. These capabilities are already valuable, and they'll undoubtedly continue to improve. But there's an important difference between tracking what a student does and understanding why they do it.
A student repeatedly asking for easier questions might be struggling with the underlying material, experiencing a loss of confidence, or simply avoiding a topic they find frustrating. A teacher who has worked with that student for several months may recognise the distinction almost instinctively, drawing on an accumulation of conversations and experiences that would be difficult to capture through an automated platform.
Sometimes the most valuable insight a teacher has comes from understanding what a student really means, even when they've said something entirely different.
Accountability and the value of personal relationships
I think another overlooked aspect of education is accountability.
Most of us have experienced how difficult it can be to sustain independent learning, whether through attempting to learn a language, completing an online course, or sticking to a revision timetable. Even when excellent resources are readily available, making consistent progress requires motivation, discipline, and a willingness to persevere through difficult material.
Teachers play an important role in developing these habits. They should be trained to establish expectations, recognise when effort begins to decline, and provide encouragement or additional structure when necessary.
There's something particularly significant about knowing that another person is genuinely invested in your progress. A teacher who remembers that you struggled with a concept last week, asks whether you've practised it, and recognises your improvement during the following lesson creates a relationship in which effort feels personally meaningful.
AI can certainly support many of these processes through progress tracking, personalised reminders, and adaptive learning plans. But I think it's worth asking whether accountability feels quite the same when it comes from a machine.
A teacher who knows a student well can recognise when they need reassurance, when they would benefit from being challenged, and when they simply need someone to encourage them to keep going. These judgements are shaped by the relationship of trust that develops over time, and can make an enormous difference to how a student approaches their education.
This is particularly important for younger students, who are still developing the independence and discipline required to manage their own learning.
The importance of the little things
One of the most interesting things about education is how much a teacher can learn through interactions that, taken individually, appear almost insignificant.
Consider a student who consistently struggles with a particular type of maths question. An AI system might identify the recurring errors, diagnose a gap in understanding, and generate targeted explanations and practice exercises. This could be incredibly effective.
But imagine a teacher observing the same student over several lessons. They notice that the student frequently arrives at the correct answer before crossing it out, seems increasingly hesitant when attempting unfamiliar problems, and occasionally makes comments suggesting that they've always considered themselves bad at maths.
Suddenly, the situation looks rather different.
The student's difficulty might have less to do with their understanding of the material and more to do with their confidence in their own ability. In this case, additional explanations and practice questions might only address part of the problem.
A good teacher can respond accordingly, creating opportunities for the student to experience success, encouraging them to explain their reasoning, and helping them develop a healthier relationship with making mistakes.
These moments happen constantly in classrooms and tutoring sessions. Teachers build a picture of their students through conversations, body language, classroom behaviour, and hundreds of small interactions that gradually provide context for the decisions they make.
AI will undoubtedly become better at recognising patterns in how students learn. However, much of the information that informs a teacher's judgement is subtle, contextual, and accumulated over time, making it difficult to imagine an entirely automated system capturing the full picture.
Confidence is a fundamental part of learning
Of all the factors that influence educational progress, I think confidence is perhaps the easiest to underestimate.
We tend to measure learning through relatively visible outcomes, such as topics mastered, questions answered correctly, and examination results achieved. These are obviously important, but they tell us surprisingly little about how a student actually experiences learning.
Consider two students with similar levels of mathematical ability. One approaches an unfamiliar problem with curiosity, expecting that it might take several attempts to find a solution. The other immediately doubts their ability, hesitates to begin, and interprets every unsuccessful attempt as confirmation that they're simply bad at maths.
Over time, these differences can become self-reinforcing. Students who believe they're capable are often more willing to tackle challenging questions, ask for help, and persist through setbacks. Those who consistently doubt themselves may gradually avoid the very opportunities that would help them improve.
This is where great teachers can have an extraordinary influence.
A teacher who has watched a student struggle with a topic for weeks can recognise the significance of a seemingly small breakthrough. They can remind the student of how far they've come, reassure them when they're frustrated, and help them develop confidence in their own ability.
And I think there's something particularly powerful about encouragement that comes from someone who genuinely knows you.
AI can provide positive feedback, celebrate achievements, and give students the opportunity to practise without the pressure of making mistakes in front of their peers. For some students, that could be enormously beneficial.
But the confidence that comes from having another person recognise your progress, understand your struggles, and genuinely believe in your potential is a valuable part of the educational experience.
Learning is an emotional journey as much as an intellectual one, and the relationships students develop with their teachers play an important role in shaping that experience.
Finding the right role for AI in education
I believe AI presents an extraordinary opportunity to improve how students learn, particularly in areas where traditional education has struggled to provide consistent, individualised support.
Imagine a classroom where every student can access personalised explanations, practise topics at their own pace, and receive immediate feedback, while their teacher has more time to focus on the individuals who need additional support. AI could help identify common misconceptions, reinforce concepts outside lessons, and reduce the considerable amount of time teachers spend preparing resources and completing routine administrative tasks.
The same opportunities exist within private tuition, where AI could provide valuable support between sessions and help students consolidate what they've learned with their tutors.
What excites me most about these possibilities is the opportunity to reconsider how teachers spend their time. If technology can take on more of the repetitive elements of teaching, we could give educators greater freedom to focus on understanding their students, developing their confidence, and providing the personal guidance that makes such a difference to the students’ progress.
However, as we explore increasingly ambitious models of AI-led education, I think we should be thoughtful about how we define success.
It's relatively straightforward to measure how accurately an AI system answers questions, how quickly it provides feedback, or how effectively it identifies gaps in a student's knowledge. It's considerably harder to measure the confidence a student develops through months of encouragement, the motivation that comes from feeling understood, or the impact of having someone recognise potential they haven't yet recognised in themselves.
These less tangible aspects of education matter enormously, even when they don't appear neatly on a progress dashboard.
There's still much to learn about how students will respond to increasingly sophisticated AI learning environments, and I expect the technology to become a valuable part of education in ways we haven't yet imagined.
But as we consider how much of teaching we can automate, I think there's an equally important question we should be asking: what aspects of education are valuable precisely because they involve another person?
The future of AI in education has enormous potential, particularly if we use it to give students access to better resources while allowing teachers to devote more attention to the people they're teaching.
Ultimately, that's the future I'd be most excited to see. One where technology makes education more accessible and personalised, while the human relationships at the heart of learning become even stronger.