In Mr Barton’s Maths Podcast (around 3:14:00), Mark McCourt shared a view that I instinctively disagreed with. He argued that technology could never replace classroom teachers because, evolutionarily, we are predisposed to value pedagogy and learning from other humans—a concept he referred to as ‘human ontogeny’ (I’d never heard this word). While this is an appealing argument (especially for teachers), it doesn’t fully convince me, not least because it presents an absolute case that human teachers will always be superior to technological platforms, regardless of the nature of the technology, the teacher and subject matter.

This perspective is at odds with the empirical reality that much of our learning now occurs without direct interaction with human teachers. Children independently explore their interests and hobbies through YouTube and books. In the workplace, most learning happens at our desks, a notable shift from the early years of my career when day-release courses were the norm for skill development. Today’s learning landscape is remarkable: we can learn almost anything we want, whenever we want, in manageable chunks that align with theories of learning and retention. This is often complemented by effective automated feedback systems that track our progress. Whether it’s through formal, paid platforms, free resources, or the eclectic mix of YouTube and online forums, the options for learning are vast and varied.

Teacher-led instruction as a motivational tool

While adults increasingly access work-related learning online, the traditional model of sending children to schools for teacher-led instruction persists. Will this approach ever evolve? Perhaps not entirely, as schools offer more than just educational support — they provide settings for socialisation, childcare, and even entertainment. However, setting these aspects aside, many remain sceptical of technology’s ability to motivate children to learn and this is the focus of this post.

For some, the argument is rooted in the belief that if left to their own devices (quite literally), young people might not engage with the subjects we deem important. For the subjects we deem important are chosen by us, not them, and they may not align with their interests. Even when students are eager to learn and achieve good grades, their adolescent brains face challenges. With executive functions still developing, they often struggle more than adults do to prioritise long-term goals, like academic success, over immediate pleasures like socialising, gaming, or smartphone use.

Teachers employ various strategies to motivate students, ranging from expectancy theory (e.g., “If you work hard for the next hour, then we can go out and play”), to empathising with the student’s situation, providing emotional support, guidance, and mentorship. They assist students with goal setting and they use explicit incentives such as house points. They sometimes strive to create an intrinsically enjoyable environment so that students find studying pleasurable, rather than a chore.

Despite these motivational strategies, there remains an often unavoidable reliance on coercion to require students to attend school and study, even where they would rather not. However, the coercive nature of schooling is not as effective in facilitating learning as one might assume. We can mandate school attendance, but we can’t ensure students’ engagement with the class material. As Graham Nuthall’s “The Hidden Lives of Learners” reveals, students in the classroom often focus on many aspects of their lives that are unrelated to the lesson.

This brings us to an important question: Can technology help us keep students on task, even in a teacher-led environment? Moreover, for secondary-aged students where independent study time is substantial, does technological innovation warrant a renewed look at its format and role?

Blurring the boundaries between teacher-led instruction and independent learning

Secondary schools have traditionally followed a daily routine of approximately five hours of teacher-led instruction, supplemented by an hour or so of independent study. Historically, we had to assume that during independent study, students would lack access to resources for looking up information, learning from instructors, or getting help when stuck. However, technology is rapidly changing this dynamic. A prime example is in mathematics, where students now have the option to learn through instructional videos and receive immediate feedback on their understanding. While this technology is not perfect, it’s evolving quickly – a glance at developments like Khanmigo offers a glimpse into the future.

The degree to which learning platforms can blur the line between instructional time and independent study varies with the nature of the subject matter. In areas particularly amenable to these platforms – especially those where practice and feedback loops are crucial – we should continually reassess not only their use in existing homework time, but also the balance between independent and teacher-led instructional time.

Many teachers are sceptical about the value of these independent learning platforms, in part because they are not present to witness the downsides of the current homework model. From a parent’s perspective, the value is clear in being able to ask a child struggling with maths: “Have you watched the video?”.

A recent article in TES featured a music teacher’s evaluation of current instrumental tuition apps, concluding their inferiority compared to traditional music lessons. While the individual instruction episode might be less effective than a weekly music lesson, the failure of most students to learn instruments is not due to poor instruction, but rather to a lack of daily practice. Music teachers might overlook this issue due to survivor bias – they themselves practised diligently as children and they disproportionately teach students who manage regular practice (since the others give up). A music app, though perhaps inferior in instructional quality, can be more effective in fostering learning if it encourages daily practice.

So, what is it about the best of these apps that makes them effective at persuading students to use them?

Learning consistent with habit-forming behaviours

These learning apps are specifically designed to encourage habit-forming behaviours. They typically promote a daily routine, leading users to adopt a consistent pattern of usage. Notifications and reminders are integral features, ensuring learners maintain this daily practice. The ease of use is a key aspect; these apps often allow users to start the day’s lesson without needing to make extensive choices or use special tools. Furthermore, many web-based learning platforms have developed mobile versions, enabling learners to engage in their studies from virtually anywhere, not just in front of a computer.

These characteristics are not inherent to all learning platforms, and they are often introduced at the expense of optimising other aspects of the learning experience. For instance, Sparx Maths doesn’t foster habit-forming behaviours and is not readily accessible for on-the-go learning, relying more on the traditional model of coercion (completing homework to avoid trouble). Sparx Maths requires a fairly large screen, at least the size of a tablet, and the use of a maths exercise book and pencil. While this might be an optimal learning environment, it doesn’t necessarily captivate the learner’s attention through habit formation as effectively as we might like it to!

Technology lacks our human flaws and needs

Mark McCourt emphasised the human teacher-student relationship as central to our learning desire. We all remember transformative teachers from our childhood who influenced our feelings about a subject, as well as those for whom we diligently completed homework, not wanting to disappoint them. But equally, some teachers failed to inspire in us any inclination to learn. Human teachers, with their diverse capabilities and personalities, have their own emotional needs. They can experience frustration and impatience with students, and sometimes respond in self-prioritising ways. Moreover, given a class of 30, providing substantial one-to-one attention to each student is a challenging feat.

Can students develop a reluctance to disappoint a non-human avatar? Although AI may never fully replicate human empathy, it can still evoke similar feelings. Duolingo’s mascot bird, Duo, sends us reminders and encouraging messages that are effective (though are likely to look very rudimentary in a decade). During the pandemic, neglect of the animals living on our family’s island in Animal Crossing induced a sense of guilt in me. As avatar animation improves, might we feel a stronger connection to them? And how will we respond when avatars appear as the voice and video of real people that we know (see HeyGen for examples)?

In the medical field, recent studies (and here) have indicated that patients often perceive AI doctors as more empathetic than their human counterparts. Is this a surprise given they can have infinite time and can be programmed to prioritise the needs of the humans they work with? In education, similar principles can be applied. By using persuasive language tailored to individual students, we could leverage AI to motivate and guide students, learning over time what type of language they respond best to.

Using personalisation to motivate learning

Although coercion is a route to motivate some learning, so much more learning in life happens as a result of curiosity, interest and necessity, and this is where the personalisation of learning to align with individual interests could be powerful if the circumstances allow it. For instance, imagine I want to learn a specific statistical technique to gain insights about teacher careers. Previously, this might have involved taking a generic online course, practising coding techniques on unrelated data like automobile sales, and reaching a high competency level before analysing one’s own data independently. Now, with the aid of large language models, I can give them my own data and converse with them to learn new techniques in the topic of interest to me. Adults who have started using ChatGPT in this way agree it is a very effective and enjoyable way to learn.

This approach is commonly used by human teachers in one-to-one tutoring settings, where learning can be tailored to individual interests, such as using football examples in math lessons or letting students choose songs for music practice or books to read. However, it becomes challenging for classroom teachers to apply this level of personalisation of examples unless all students share similar interests and motivations. AI-supported independent learning platforms present a potential solution, offering personalised examples or learning pathways that align with a student’s interests or areas of existing knowledge.

Of course, there are limitations to how far key curriculum concepts can be tailored to student interests in some subjects. For instance, it’s hard to imagine teaching cell structures or the rise of the Nazis with football examples! The feasibility of this approach largely depends on the specific knowledge architecture of each subject.

Learning how technology and humans can work together to motivate students

Do I believe AI can be as motivating, or even more so, than the average teacher? It’s possible in the future. Teachers vary a great deal and their motivational prowess isn’t always strong enough to compete with the allure of spending time with friends or staying home. For now, I doubt AI will completely replace the traditional school setting that enforces daily learning, regardless of the motivations of the student. I do, however, suspect the future of education will leverage the strengths of technology, materially changing the structure of lessons and independent learning, while retaining the irreplaceable human elements of mentorship and emotional support.