Adaptive Learning Systems

An adaptive learning system changes what a learner sees according to evidence of that learner's current knowledge and performance. It uses a continuously updated learner model to adjust content, difficulty, pace, feedback and support, rather than giving every learner an identical sequence.

The data-feedback cycle

Personalisation operates as a repeated cycle of measurement, inference, selection and reassessment.

• The system collects performance data such as answer accuracy, response time, number of attempts, skipped items and use of hints.
• Diagnostic questions and ongoing formative assessment help estimate which concepts the learner has mastered and where misconceptions or gaps remain.
• Rules or algorithms update the learner model and select the next suitable activity.
• The learner's response supplies fresh evidence, so the pathway is revised continuously rather than fixed permanently.

What the system personalises

Adaptation can occur at several levels, from the next question to the broader learning pathway.

• A learner showing mastery may receive harder tasks or move ahead, while another may receive remedial content, worked examples or prerequisite material.
• The system may alter the sequence, pace, amount of practice and form of feedback.
• Content may be presented through different formats, but meaningful adaptation should follow demonstrated learning needs rather than superficial preferences.
• Dashboards can provide teachers with evidence for targeted instruction, grouping and additional support; the system complements rather than replaces professional judgement.

Educational value and safeguards

Frequent feedback can support mastery learning and help address differences within a classroom. Its effectiveness, however, depends on sound pedagogy, valid assessment and equitable access.

• Limited or poorly designed data can misclassify learners and repeatedly direct them into unsuitable pathways.
• Algorithms and content require checks for bias, transparency and accessibility across languages, disabilities and social contexts.
• Collection of children's data requires privacy, security, informed governance and data minimisation.
• Human oversight is necessary because motivation, creativity, socio-emotional needs and classroom context cannot be inferred reliably from performance scores alone.