Personalisation is often discussed as an educational good. More relevant resources, more flexible pathways, more responsive support. The assumption that sits behind many of these discussions is that increasing personalisation improves learning.
These interventions are usually justified on educational grounds. They help learners find appropriate resources, reduce navigational burden, and focus attention on activities that matter. Personalisation often works exactly as intended. What is not intended is the more interesting question: how personalisation influences learner behaviour and what happens when support begins to shape decisions rather than simply inform them.
In a previous post, I explored whether more choice necessarily creates better learning experiences. Personalisation introduces a related challenge. Even when choices remain available, systems can influence which options are seen, prioritised, and ultimately selected.
The Problem with Neutrality
Personalisation is often described as a neutral process. The system identifies relevant information and presents it to the learner in a more useful form based on data rather than the educator judgment. This creates an illusion of neutrality and impartiality, yet personalisation is never entirely neutral.
Every personalised recommendation reflects assumptions about what should be prioritised, what success looks like, what information is relevant, and how learners are expected to interact with whatever is presented. A recommendation to revisit content assumes revision is beneficial or a reminder to complete an activity assumes that activity deserves attention now.
Even when these assumptions draw on solid evidence and professional experience, they remain assumptions. The moment a system begins influencing decisions, questions of judgement and responsibility become difficult to avoid. These questions become even more difficult when recommendation systems are powered by AI. The learner may not understand why particular resources, pathways, or activities are being suggested. Educators may not fully understand either. As recommendations become more complex, the assumptions shaping them can become harder to identify and challenge. The distinction between supporting and shaping decisions can become harder to see as well.
Supporting Decisions and Shaping Decisions
Not all forms of influence are equally problematic and most learning design involves some degree of persuasion. Learning outcomes direct attention towards particular goals, activity sequences encourage specific behaviours, and assessment structures reward some forms of participation while discouraging others.
Learning has never been neutral, so attempting to eliminate influence is a nonstarter. The more useful distinction may be between supporting decisions and shaping them. In practice, these can be difficult to separate but supporting decisions means helping learners understand their options and consequences whereas shaping decisions means making certain outcomes significantly more likely than others.
However, a recommendation may feel like information to one learner and instruction to another and a default setting may be experienced as the expected choice even when alternatives remain available. The influence exists regardless of whether it was intended.
When Convenience Becomes Direction
One reason personalisation is attractive is that it reduces cognitive load. Learners do not need to search through large volumes of content, compare every available option, or determine an optimal pathway from scratch. The system helps organise complexity. That convenience can be valuable, but it can also make influence difficult to see.
It is difficult to determine, for example, whether a recommended option is selected because it is genuinely the best choice or simply because it is the easiest to access. Another example would be distinguishing between preference and compliance if learners consistently follow a suggested route.
As systems become more sophisticated, it will become easier to guide behaviour while concurrently becoming more difficult to distinguish between learner choice and system influence. Evidence will need to be interrogated even more thoroughly to distinguish quality, visibility, compliance, and ease when drawing conclusions.
The Hidden Role of Defaults
Defaults rarely attract attention because they are often experienced as normal, which perhaps makes them the most powerful form of persuasion. They are simply what is. The default pathway, the pre-selected option, the recommended resource. Defaults appear to be administrative rather than educational and yet they shape behaviour precisely because learners accept them without a second thought. Constantly evaluating every available option would be exhausting so it is not irrational. The problem is that defaults often embed decisions that remain invisible to learners.
What appears natural may be the result of a series of earlier design judgements that decided, and privileged, one pathway, one set of recommendations, and the fulfilment criteria. The learner encounters the outcome without necessarily seeing the reasoning and assumptions behind it.
The Question of Autonomy
This is where discussions about autonomy become more complex because autonomy is often interpreted as the presence of choice. By that definition, a learner with ten options possesses more autonomy than a learner with two. But it simply isn’t that straightforward. Unlimited choice can create confusion, uncertainty, and navigational burden. But excessive guidance can reduce meaningful decision-making. Both situations can undermine autonomy in different ways.
The challenge becomes particularly visible when learner variability is taken seriously. Choices that support one learner may create navigational or cognitive burden for another.
Increasing the number of available options doesn’t automatically create meaningful autonomy. Learners must understand their choices, navigate them with confidence, and make decisions that remain genuinely their own. A learner overwhelmed by possibilities may suffer from decision fatigue and exercise less autonomy than a learner operating within a well-designed structure. Equally, a learner who follows recommendations without understanding alternatives may appear autonomous while experiencing very little meaningful control.
The challenge is creating conditions in which learners can exercise judgement rather than just maximising freedom.
The Limits of Personalisation
Systems work by identifying patterns, categories, indicators, and characteristics that can be acted upon. Real human beings rarely fit neatly within those categories and so every form of personalisation simplifies differences in order to function. Therefore, personalisation cannot fully live up to the promise of accommodating individual differences even if the factors influencing learning existed in a form stable enough to be fully captured through data.
This does not make personalisation ineffective, but it does suggest caution about claims that systems can reliably determine what learners need better than learners themselves. Understanding the limits of personalisation makes it easier to recognise where it remains useful and where its claims begin to exceed what can reasonably be supported.
Beyond Personalisation
Systems have always influenced behaviour and framing the discussion purely in terms of personalisation can obscure more important questions. How visible that influence is, how transparently it operates, and who remains responsible for the consequences.
The ethical significance of personalisation does not emerge from the technology itself. It emerges from decisions about whose interests are prioritised, which assumptions remain visible, and who bears the consequences when those assumptions fail to hold. Support becomes problematic when it becomes difficult to distinguish from direction and guidance becomes problematic when its assumptions become invisible. Personalisation becomes problematic when learners begin adapting themselves to systems rather than systems supporting learners.
These arguments are a reminder that learning design choices are fundamentally about ethics, influence, responsibility, and the conditions under which learners exercise autonomy, not just technical capability.
Those decisions deserve scrutiny not because personalisation is uniquely risky, but because they are often easier to see once personalisation brings them into view.
Leave a Reply