Human Centred Learning

By Mark Butler PhD

AI Enabled Adaptive eLearning

Imagine a brilliant teacher who knows all her students’ interests, skills and preferred ways of learning. She continuously learns about each student, has the luxury of time to give individual support, and checks for understanding to decide whether to teach further or move onto the next topic. She tracks the progress of each student and uses this information to change how she teaches (differentiated teaching).

This is an analogy for adaptive learning – a process of understanding learners and adapting the design and delivery of courses to best suit them. For eLearning, adaptive learning can be achieved at scale through emerging artificial intelligence (AI) tools.

Adaptive eLearning assesses learner needs to dynamically tailor curriculum to meet those needs. A Learner Profile of existing information about the learner alongside tracked behaviours interacting with curriculum changes the sequence of curriculum, the way content is presented, the pace, the duration, and the recommended future courses. Data can now be used immediately to create new variations of a course while the learner takes that course.

AI has already enhanced many course catalogues by suggesting learning based on role, goals and past course completions. Generative artificial intelligence (AI) can take this further by analysing learner data and tailoring the curriculum delivered. This involves building and rebuilding curriculum and monitoring how a learner reacts. This means learners will learn in ways best suited to their changing needs. By linking learning analytics with generative AI, we move from curation of existing content to near-instant creation of new content.

We now have tools that quickly make decisions to design and redesign modules that:

Build a learning journey of courses and resources then rebuild this learning journey as skills improve or interests change.
Skip skills and knowledge that the learner demonstrates they have already.
Scaffold topics by repeating complex topics not fully understood and re-explaining these topics more simply.
Hide learning activities from all learners that are not effective in improving understanding.
Add learning activities that meet skills gaps or respond to learner interests.
Replace learning activities to change the mix of visuals, text, audio, scenarios, simulations and video.
Sequence topics to change the order based on learner performance and interest.
Predict what skills should be prioritised next and recommend resources, coaching opportunities and live training.
Augment instructor-led sessions through quizzes to learners that provide feedback to instructors during delivery.

These actions all customise the learning experience to stop pushing out one-size-fits-all content. Learners might still choose some of the topics and activities, but this approach goes beyond preferences to learn what works best for the learner.

What’s Possible with the Right AI Tools?

Adaptive eLearning can analyse learner needs, skill sets, learning preferences, and cognitive abilities of each learner. This can occur through historical learning data, pre-course assessment and monitoring of how learners interact with content. A system can experiment by presenting content in different modes such as text and video then check which content a learner preferred, and which content led to better quiz results or scenario choices. Through modular topics and microlearning, personalisation of learning journeys can be configured immediately.

To tailor eLearning, a Learner Profile can start with the learner’s role and historical learning data. Then, that profile is constantly updated depending on how the learner interacts with learning activities. The following examples show how adaptive eLearning can automate a cycle of analysis, design, and redesign.

Analysis	Design	Redesign
Alec is a new hire with a decade of experience in sales. He needs to complete the selling skills course but is likely to only lack skills in company-specific procedures. The platform analyses Alec’s CV, role description and initial skills assessment.	A customised eLearning course is built with only 3 of the 12 selling skills modules enabled as these focus on company policy and procedure.	A quiz in topic 2 identifies a need for Alec to understand recent compliance changes so an extra module is immediately added to his learning journey.
A return-to-work plan is created for Jenna to reintegrate her back to her team. Jenna is on reduced hours and for now, she can only concentrate for short periods.	Jenna sees a catalogue of microlearning relevant to her role. She slowly progresses in each course and gets some quiz responses incorrect. The system reconfigures the content to include multiple explanations of the same concept.	Jenna begins to move more quickly through courses and quizzes so more complex case scenarios are delivered and longer courses are recommended. A digital coach is embedded to courses to encourage and check for understanding.
Slava resists completing his eLearning and believes he already has all relevant skills. He is recommended a course that starts with a quiz which gets progressively harder.	With strong quiz results answered quickly, a short simulation course is recommended and adjusted to be more complex.	Slava is recommended other more advanced microlearning courses and simulations.
A mandatory organisation-wide course is planned for blended delivery. Individual needs differ greatly across divisions and roles. An eLearning course before the instructor-led session creates individual Learner Profiles.	Learner skills gaps are used to recommend instructor-led session sessions with in-depth sessions recommended for those with pressing needs.	Instructors in live sessions receive a summary of the Learning Profiles for all learners attending.

ADDIE Deconstructed

The ADDIE model of instructional design gets pulled apart byAI-enabled adaptive eLearning. The phases we know so well of Analyse, Design,Develop, Implement, and Evaluate remain relevant but do not occur sequentially.

AI-enabled adaptive eLearning instead analyses and evaluates constantly and then automatically adjusts the design in real-time during delivery. Instructional design becomes much more complex in determining what core topics and approaches must remain consistent to achieve learning outcomes and what can be tailored depending on each Learner Profile.

For adaptive learning, we might consider amore iterative model of ADDER: Analyse, Design, Develop, Evaluate and Redevelop. The implementation phase here is constant and automated with iterations of the eLearning constantly tested and refined.

With so many variables being monitored and reacted to, several steps are repeated like the analysis step of understanding ‘who your learners are’ and the evaluation step of ‘what changes can improve the course’. At scale, only AI-enabled systems can properly determine which variables matter most and what types of changes have the best learning outcomes.

ADDER
Analyse	Design	Develop	Evaluate	Redevelop

Tools and Platforms

AI-enabled adaptive eLearning may become a feature of popular Learning Management Systems (LMS) and Learning Experience Platforms(LXP). To go beyond what could already be done with machine learning though, we need better authoring tools. Popular authoring tools like Articulate 360 already have generative AI but cannot adapt to learner needs dynamically. More likely we are going to see AI-enabled adaptive eLearning available only in platforms that are fully integrated LMS, LXP and authoring.

Platforms focused on creating a standalone solution using adaptive learning include Sana Learn, Realizeit, Disco, Otto Learn, Area 9 Lyceum, SCTraining, Cornerstone and Open LMS. There are also tools like Adaptemy andProLearn that aim to integrate with existing systems to profile and personalise learning.

These solutions mostly personalise learning journeys or make small customisations to learning content. Through the Experience API technical standard (xAPI), we are also learning detailed information about how a learner interacts with online course activities and offline training activities.Adaptive learning based on xAPI can already identify where a learner is struggling and add in additional learning activities.

Current tools have not yet realised the dream of mass personalisation of courses. So why should we care about AI-enabled adaptive eLearning right now? These tools will arrive as features or add-ins to existing tools or as fully integrated standalone platforms. When they do, we should prepare by:

Collecting data on detailed learner interactions with eLearning through advanced data collection via xAPI and other learning analytics.
Creating further microlearning courses and more modular sections to be usable as moveable chunks in learning journeys.
Asking your LMS or LXP vendor about their roadmap for including AI tools and adaptive learning.
Deciding whether your organisation might train a custom generative AI tool on your existing course library to dynamically build new adaptive eLearning.
Identifying key groups of learners who most need a new approach to eLearning such as your most time-poor staff.

Towards a Pilot of Adaptive eLearning

Adaptive learning changes instructional design and creates anew complexity to how we design and deliver eLearning. There are however some groups of learners who have been particularly difficult to engage in eLearning. A pilot program testing adaptive eLearning with harder to reach groups could be a first step in this human-centred personalisation.

Adaptive learning promises to create learning experiences where learners feel more engaged and confident to apply new skills. Learning experiences that constantly adapt to better resonate with learners have significant potential and generate a lot of data that would be useful to evaluate a pilot.

Through implementation of these tools, we have scalable ways for learners to ‘choose their own adventure’. Learners can be given choices while also experiencing curriculum that is optimised in terms of how long it takes for them to become competent and knowledgeable. For organisations, adaptive learning reduces the time spent on training, makes eLearning more impactful, identifies skills gaps and quickly measures the proficiency of staff.

Interested in AI and eLearning?

AITD offers several courses on both AI and eLearning:

AI Essentials for L&D Professionals: Do you want to transform your L&D workflow and enhance learner experiences using Generative AI? In this blended learning course, you’ll learn how to use Generative (Gen) AI for a range of key L&D functions, including skills gap analysis, content creation, personalised learning and feedback. Register now.

eLearning: Foundations: This is Part I of an engaging, social learning suite of courses that provides you with access to learning experiences, activities and a comprehensive knowledge base. Register now. You may also be interested in eLearning: Planning and Design; and eLearning: Production and Delivery.

About the Author: Mark Butler PhD

Mark Butler PhD is a Learning Consultant working in the public sector in diversity and inclusion training. Mark completed his PhD in education management and has published on higher education, wellbeing, and diversity and inclusion. Mark has worked across corporate, government and nonprofit sectors with a focus on eLearning and facilitation of workshops that encourage a change in attitudes and behaviours.

Access