Gerald Ajam

“Personalisation is inherently learner-centred”

Personalisation serves learners when it adapts to what they know, not to what they prefer.

Status Position

Published September 2026

All counterpositionsLearning

Usually said

“Personalisation is inherently learner-centred”

The shift

From: Adapting to the learner

To: Adapting to what the learner knows

More precisely

Personalisation is learner-centred when it adapts guidance and practice to what each learner currently knows and keeps the teacher able to see and override it.

In short

Adapting teaching to each learner can help a great deal, if the system adapts to the right thing. Adapting to what a learner already knows is well supported. Adapting to what they prefer or enjoy is not, and can quietly narrow what they learn. Who controls the adapting matters too.

The argument

From the claim to a better one

  1. The strongest formulation

    Learners differ in what they already know and how fast they progress. A system that adjusts pace, sequence and support to each learner wastes less of anyone’s time than one lesson pitched at the middle.

  2. Why it is attractive

    It promises one-to-one tutoring at the cost of software. It also sounds respectful: who would argue for treating every learner the same?

  3. Where it holds

    It holds where adaptation tracks prior knowledge and performance: more guidance for novices, less as they improve, and practice targeted at what has not yet been learned.

    Where it breaks

    It breaks when it adapts to preference, style or engagement, serving more of what the learner already likes or does well. It can also narrow the curriculum, isolate learners from shared discussion, and move decisions from teachers to a model nobody can inspect. Part of the disagreement is evidence; part is values, about who should decide what a learner does next.

  4. What evidence matters

    Kalyuga et al. (2003) on the expertise reversal effect support adapting guidance to prior knowledge. Pashler et al. (2008) found no adequate evidence for matching instruction to learning styles. Pane et al. (2015) at RAND found modest and uneven gains from personalised learning, with implementation varying widely between schools. What matters is what is being adapted, to what signal, and with what effect on learning rather than use.

  5. So, more precisely

    Personalisation is learner-centred when it adapts guidance and practice to what each learner currently knows and keeps the teacher able to see and override it.

Reading it

The precise version, word by word

Personalisation is learner-centred when it adapts guidance and practice1 to what each learner currently knows2 and keeps the teacher able to see and override it3.

  1. Guidance and practice

    How much support a learner gets, such as worked examples, hints and structure, and what they practise next.

  2. What each learner currently knows

    Their prior knowledge and recent performance, as shown by what they can do. Stated preferences, supposed learning styles and time spent are weaker signals.

  3. See and override it

    The teacher can tell why the system chose what it chose, and can change it. A decision nobody can inspect has left the classroom.

What it changes in practice

  • Ask any personalised product what it adapts, to which signal, and what evidence links that to learning.
  • Fade support as learners improve. Novices need more guidance, and experts can be slowed down by it.
  • Keep shared work and discussion in the week, so personalised practice does not turn into learning alone.
  • Make the system’s choices visible to teachers, with a way to change them.

A case, in general terms

Two maths platforms both call themselves personalised. One notices that a learner keeps getting fraction division wrong, gives more worked examples, then fades them as accuracy improves. The other notices that the learner enjoys geometry puzzles and serves more of them. Both are personal. Only the first adapts to what the learner needs to learn next.

Singapore