For a long time, learning has revolved around one central figure: the teacher. And understandably so, because we have all had teachers who helped us understand things, answered our questions or introduced us to new ideas. However, I am not sure that this premise will hold true in the future.
I understand why this idea might feel uncomfortable. Those of us who work in training know the value of a good teacher and have good reason to defend it. But we should also be able to ask ourselves how much of that value depends on capabilities that artificial intelligence could eventually develop.
My view is that artificial intelligence will evolve to take on many of the teaching functions currently performed by teachers, tutors and trainers. I am not talking about the present or the next few years, but about a gradual transformation that could completely redefine how we learn and who guides us through that process.
Before discussing whether artificial intelligence could replace a teacher, though, I would like to start with a simpler question: how would we want to learn if we could choose the best possible way?
Learning that adapts to each person
Imagine having someone who knows our level, understands where we get stuck and finds a way to explain things that helps us make sense of them. If we progress quickly, they increase the difficulty. If we need to pause, they revisit the problem without making us keep pace with everyone else.
That support would also take our interests into account. It could explain a concept using examples relevant to our work or a hobby, recognise what makes us lose motivation and help us get back on track. And it would retain what it had learnt about us when we started another course, changed careers or wanted to develop a new skill.
It is hard to compare that experience with a model in which everyone receives exactly the same explanations and follows the same timetable. When that happens, there are always people who find the pace too fast and others who find it unnecessarily slow.
Teachers try to respond to these different needs every day, but there are clear limits to the time available. Providing such individualised support for every learner, consistently over the years, would require enormous resources.
That is why I find AI in learning so interesting. Not because it replaces the value of education, but because it could make possible a level of personalised support that has, until now, been very difficult to offer on a large scale.
A tutor that gets to know us
What I have in mind is a tutor that, as well as teaching us, also learns how to help us. One that remembers our difficulties and checks whether its explanations work. One that can suggest a different activity when repeating the same thing is not getting us anywhere.
Of course, reaching that point involves much more than building an AI that can hold a conversation. It would need to work with reliable information, interpret each person’s progress correctly and recognise when it is not achieving the intended result. Its success would be measured not by the quality of its answers, but by the learning it brings about.
I believe we are heading in that direction and moving towards that model, although we are still a long way from seeing it fully developed. And it will probably take very different forms from those we see today. Perhaps we will interact through voice, devices that accompany us throughout the day or technologies we are only just beginning to imagine.
If an AI could support us in that way, with good results and at an affordable cost, the transformation would affect the entire education model. We could support far more people with much smaller teams. That would mean rethinking the role of teachers, the purpose of learning institutions and the role of those of us who design technology to support learning.
Technology and the sector move at different speeds
The history of technology shows that technical feasibility and widespread adoption rarely move in step. I think we will have to navigate three different paces here: the pace of AI development, the pace at which those of us working in the sector can integrate it into our work, and the pace at which learners adapt. Just because a tool can solve a problem does not mean we will start using it the next day.
For a training provider, making the most of these advances may mean reorganising its courses. For a teacher, rethinking certain tasks. And for a learner, getting used to a different way of learning. These changes take time too.
There is also a clear cultural barrier. Many people will continue to place greater trust in familiar models than in new alternatives, even when those alternatives offer objective advantages. Acceptance of change depends as much on perceptions and habits as on the effectiveness of the technology itself.
Then there is the financial question: individuals and organisations will need to see clear value in the tools they pay for. I see considerable potential in training, because giving each learner individual attention usually requires more time and resources than can readily be provided on a large scale.
Added to all this is regulation. For example, in Spain’s subsidised online training system, FUNDAE’s requirements include tutoring, progress monitoring and interaction between tutors and participants. New models will have to fit within that framework, and we will need to discuss how the framework itself should evolve to recognise other forms of support while maintaining safeguards.
If tools become available that can provide this support effectively, I suspect that adapting the way we work and learn will take longer than introducing the technology itself. These tools may be ready to teach us before we feel ready to learn with them.
Teachers as artisans of education
If we look several decades ahead, a possibility emerges that those of us in the sector may find uncomfortable: training delivered by people could cease to be the predominant model.
Take craftsmanship, for example. We still value a handmade piece of furniture or a garment made by a professional for a particular person. We might choose that work for its quality, for the direct relationship with its maker or because we value a more personal experience.
Could something similar happen with learning? Could there come a time when learning with a human teacher is something we choose for the particular value it offers, while learning directly with AI becomes the norm?
That does not strike me as far-fetched. And that is precisely why I think this debate is worth having. Taking our role for granted could prevent us from looking clearly at the distinctive contribution we make and how it might evolve over the coming years.
After all, trusting that people will continue to play a meaningful role is one thing. Assuming they will play exactly the same role in twenty or thirty years’ time is quite another.
Human connection will still have a place
There is an obvious objection: learning also brings us into contact with other people. We share experiences, listen to different opinions and discover other ways of understanding the world. For many learners, that human element is just as valuable as the learning itself.
That is why, when we talk about bringing AI into learning, we also need to consider how to nurture opportunities for conversation, connection and shared learning. The question is what place we give them within these new approaches to learning.
It is easy to imagine scenarios in which both dimensions coexist. We could have time for individual learning alongside opportunities for conversation, collaborative work or sharing ideas. A learner could explore a topic with their AI tutor and then discuss it with other people. Or practise a skill independently and later take part in an activity where interaction with others matters.
The balance would need to suit each context. Vocational training for adults, a childhood learning experience and a course someone attends both to learn and to meet people all involve different needs. Preserving the value of human connection means understanding those differences.
Even in highly automated settings, there will still be important decisions to make: which content to prioritise, how to assess outcomes or what to do when the tools do not work as expected. What is less clear is whether those responsibilities will be shared out in exactly the same way as they are today.
Staying one step ahead
This whole debate has a very practical implication for those of us who work in training: we need to start questioning what we do, why we do it and whether doing it the same way will still make sense in a few years’ time.
Content creation is a good example. If AI can quickly produce a first draft of a presentation, an exercise or an assessment, our work changes straight away. The focus shifts from producing materials to checking them, improving them and ensuring that what we design actually helps people learn.
That raises an uncomfortable but necessary question. If technology performs a task better than we do, are we willing to let it? In many cases, the answer will have to be yes, even if that means reshaping responsibilities we currently take for granted as part of our work.
Of course, there is no path without mistakes. Some decisions will prove right and others will not. Some innovations may also arrive before learners and organisations are ready to adopt them. But standing still does not remove the risk either. In fact, it could mean that, by the time we decide to act, others have already covered much of the ground.
It therefore seems more sensible to move forward through short learning cycles: test, measure the results and adjust what is not working. The guiding principle should always be the same: check whether learners are learning better.
This also has implications for us as technology companies. If the ways people learn evolve, we will need to keep reviewing the role our solutions play and the problems they actually help solve.
I believe AI will eventually teach much of what people teach today. I do not know when that will happen or how far the change will go, but I think it is a scenario worth preparing for. At evolMind, we too will need to reassess the value our tools bring.
If we find that a learner learns better with AI, we will need to be willing to let it take on tasks we currently perform ourselves, even if that changes the way we work. Accepting that is also part of our responsibility to those who learn.