Skip to content

Lots of blogs are declaring the ‘new’ trends in e-learning for 2022. There is nothing truly new in most of these. The difficulty with trends is they frequently draw from the ‘thought-osphere’, rate than from emergent practice. One of these trends predicted for 2022 is ‘adaptive learning’.

banner image of road network at night

This ‘trend’ has been around for well over a decade and still cannot gain significant traction. Few would disagree that learning should be personalised to the largest degree possible to ensure motivation and relevance for the learner. Adaptive learning is the notion that we can tailor learning for the individual based on past performance and the ultimate direction. A little like the way an intelligent GPS navigation system for your car can suggest alternative routes when you take a wrong turn.

Educators have long implemented adaptive learning strategies in face-to-face contexts, usually as differentiated teaching practices. Teachers will tell you that for this strategy to be effective, class sizes need to be small and that you need to know your students at a human level. Knowing not just that they struggle with numerical concepts, but also that they enjoy music, or sports, so that we can pose exemplars as problems in a language that is meaningful to that learner. Differentiated teaching is highly personalised.

In contrast, adaptive learning systems, which are based on the interactions individuals have with computer based learning activities, simply cannot consider the societal influences, the personal likes and dislikes, of the individual. Sophisticated adaptive systems allow the learner to be presented with alternative phrasing of a problem if they appear to misunderstand instructions, or to be presented them with simpler versions of tasks if the initial one appears to be too difficult.

Adaptive systems are limited to the domain of knowledge acquisition, learning stuff, and some lower cognitive skill development. Essentially, adaptive learning is an old-style computer-based training (CBT) course on steroids. Facts and cognitive processes are important, and delivering up examples that are appropriately positioned to stretch the learner is a good thing. We should all understand, however, that the technology lags way behind the aspirations of teachers. It is a far cry from personalising learning.

Personalised learning is best enabled by situating learning experiences within the social context in which the learner lives and works. Allowing the learner to design their own responses to assessment tasks, to generate personalised evidence of their learning and relate their learning to their real lives. True personalised learning is essential for four domains of educational objectives, affective, psychomotor, interpersonal and the intrapersonal (meta-cognitive), for the factual knowledge ‘stuff’ and some cognitive skills there are, or may be, adaptive learning systems.

Unfortunately for budget-holders and institutional leaders, personalised learning requires a staff-student ratio that defies current budgets and, my particular interest, carefully crafted curriculum, programme and course design. For some, it appears to be worth investing in ‘automated’ learning systems, sold on the promise of responding directly to the student’s needs. Systems hyped by the vendors frequently underrepresented the investment needed in designing alternative branching scenarios and associated questions. Most vendors promise banks of questions based on relatively simple algorithms. Until computing power is significantly increased, answers and questions can be truly automated through AI systems, and systems can draw accurate student profiles based on social media and shared data (a worrying possibility for many) adaptive systems will remain a limited tool for specific contexts.

These contexts include mathematics and most ‘hard’ sciences. Where there is a required base of factual knowledge that is widely regarded as uncontested, adaptive learning systems can provide a marginally more engaging version of rote learning. It may even provide some ability to prompt the learner to transfer knowledge from one context to another beyond pure memorisation. I contest its applicability is still limited to the cognitive domain.

Still, the hyping of adaptive systems continues and they remain on most 2022 trends list. Clearly, one trend that I confidently predict for 2022 is that technological determinism, the concept that technology is intimately related to our social development, will continue to feature in the ‘trends’ blogosphere.

Photo by Ishan @seefromthesky on Unsplash

12274589_10153905502716282_5140343966725134468_n.jpg

Atkinson, S.P. (2015). Using Learning Design to Unleash the Power of Learning Analytics. In T. Reiners, B.R. von Konsky, D. Gibson, V. Chang, L. Irving, & K. Clarke (Eds.), Globally connected, digitally enabled. Proceedings ascilite 2015 in Perth (pp. 358-364 / CP:6-CP:10).


 

A very enjoyable presentation made this week at ascilite 2015 in Perth, Australia. Wonderful to engage with this vibrant and hospitable community. Amongst some fascinating presentations exploring the theoretical and information management dimension of learning analytics and academic analytics, my very foundational work on constructively aligned curricula and transparency in design was I believe welcomed.

I said in my paper that I believed "New learning technologies require designers and faculty to take a fresh approach to the design of the learner experience. Adaptive learning, and responsive and predicative learning systems, are emerging with advances in learning analytics. This process of collecting, measuring, analysing and reporting data has the intention of optimising the student learning experience itself and/or the environment in which the experience of learning occurs... it is suggested here that no matter how sophisticated the learning analytics platforms, algorithms and user interfaces may become, it is the fundamentals of the learning design, exercised by individual learning designers and faculty, that will ensure that technology solutions will deliver significant and sustainable benefits. This paper argues that effective learning analytics is contingent on well structured and effectively mapped learning designs."

12316100_10153905502676282_6800926538129710150_n

Back in the late northern hemisphere summer of 2013 I drafted a background paper on the differences between Educational Data Mining, Academic Analytics and Learning Analytics. Entitled 'Adaptive Learning and Learning Analytics: a new design paradigm', It was intended to 'get everyone on the same page' as many people at my University, from very different roles, responsibilities and perspectives, had something to say about 'analytics'. Unfortunately for me I then had nearly a years absence through ill-health and I came back to an equally obfuscated landscape of debate and deliberation. So I opted to finish the paper.

I don't claim to be an expert on learning analytics, but I do know something about learning design, about teaching on-line and about adapting learning delivery and contexts to suit different individual needs. The paper outlines some of the social implications of big data collection. It looks to find useful definitions for the various fields of enquiry concerned with collecting and making something useful with learner data to enrich the learning process. It then suggest some of the challenges that such data collection involves (decontextualisation and privacy) and the opportunity it represents (self-directed learning and the SOLE Model). Finally it explores the impact of learning analytics on learning design and suggests why we need to re-examine the granularity of our learning designs.

I conclude;

Learning Analytics Cover"The influences on the learner that lay beyond the control of the learning provider, employer or indeed the individual themselves, are extremely diverse. Behaviours in social media may not be reflected in work contexts, and patterns of learning in one discipline or field of experience may not be effective in another. The only possible solution to the fragmentation and intricacy of our identities is to have more, and more interconnected, data and that poses a significant problem.

Privacy issues are likely to provide a natural break on the innovation of learning analytics. Individuals may not feel that there is sufficient value to them personally to reveal significant information about themselves to data collectors outside the immediate learning experience and that information may simply be inadequate to make effective adaptive decisions. Indeed, the value of the personal data associated with the learning analytics platforms emerging may soon see a two tier pricing arrangement whereby a student pays a lower fee if they engage fully in the data gathering process, providing the learning provider with social and personal data, as well as their learning activity, and higher fees for those that wish to opt-out of the ‘data immersion’.

However sophisticated the learning analytics platforms, algorithms and user interfaces become in the next few years, it is the fundamentals of the learning design process which will ensure that learning providers do not need to ‘re-tool’ every 12 months as technology advances and that the optimum benefit for the learner is achieved. Much of the current commercial effort, informed by ‘big data’ and ‘every-click-counts’ models of Internet application development, is largely devoid of any educational understanding. There are rich veins of academic traditional and practice in anthropology, sociology and psychology, in particular, that can usefully inform enquiries into discourse analysis, social network analysis, motivation, empathy and sentiment study, predictive modelling and visualisation and engagement and adaptive uses of semantic content (Siemens, 2012). It is the scholarship and research informed learning design itself, grounded in meaningful pedagogical and andragogical theories of learning that will ensure that technology solutions deliver significant and sustainable benefits.

To consciously misparaphrase American satirist Tom Lehrer, learning analytics and adaptive learning platforms are “like sewers, you only get out of them, what you put into them’."

Download the paper here, at AcademiaEdu or ResearchGate

Siemens, G. (2012). Learning analytics: envisioning a research discipline and a domain of practice. In Proceedings of the 2nd International Conference on Learning Analytics and Knowledge (pp. 4–8). New York, NY, USA: ACM. doi:10.1145/2330601.2330605

%d bloggers like this: