L&D is always looking to improve; it’s in the name after all. This applies to efforts ranging from improving technology via improved learning design to a variety of in person, hybrid, and digital strategies: Push, pull, nudge, learning in-the-flow-of-work, and social or peer-to-peer learning, to name only a few of the buzzwords we encounter time and time again, across the industry.

These improvements do indeed tend to move the dial in the right direction for learning and performance. They improve the experience generally, increase the accessibility of resources and courses specifically, and accomplish all of this from a variety of different angles, pathways, or recommendations. However, we also need to take stock and acknowledge that, albeit good steps, these are merely small steps in the right direction.

How come?

Because the most important factor for impactful and effective learning is true personalization and feedback, during and throughout the learning journey.

Over-reliance on one-size-fits-all training

We’ve known this for a long time, and in particular from Benjamin Bloom’s research piece on personalized learning and feedback, “The 2 Sigma Problem.” Regardless of the aforementioned efforts, what we are delivering is still a one-size-fits-all solution, albeit in a range of guises.

The one-size-fits-all approach is not limited to digital training and LXP efforts. When we look closely at the world of training and performance improvement, we encounter this shortfall, this lack of true personalization, in a range of learning formats, whether workshops or training courses, where instructors use the same slides and resources over and over for completely different cohorts of learners, in very different settings. Despite offering in-person training and learning, these trainers are still stuck in a one-size-fits-all approach.

This is now changing.

AI aids in providing true personalization

Automation through AI is providing us with the tools necessary to deploy fully personalized digital learning, extremely fast and at scale. With the advent of this technology, we will see a revolution in digital training; in addition, I predict that the impact the digital piece will have on human-led efforts will lead to a further revolution of education, training, workshops, mentoring, and coaching.

The launchpad for these groundbreaking changes is AI contextualization, which we explored in “Boost Usability of Libraries & Knowledge Hubs with Automation.”

AI contextualization engines are able to map out all forms of digital content into different concepts and align them with learning objectives, capabilities, skills, business verticals or horizontals, or any other framework. In addition to being able to contextualize and "organize" content through mappings, the AI engine also automatically builds the complexity tree of the topics and concepts covered, as well as whether and how they are related. All of this is very similar to how a good textbook would use instructional scaffolding to guide the learning journey, starting from the least complex chapters, requiring little or no prerequisites, to the most complex concepts, relying on previous chapters to have been covered.

Why is this key?

Automating the ‘personal tutor’

To understand this better, we need to take a look at how a personal tutor would successfully personalize learning for their student. First and foremost, such a tutor would possess excellent subject knowledge, say for example in eighth-grade math. This means that they have understood the different concepts and topics covered, how each fits into the curriculum, and how the different topics and concepts are related to one another, for instance how they build on one another in terms of required prior knowledge. Furthermore, the tutor would have a good understanding of how the different math concepts need to be applied within the scope of the curriculum, the methods of assessment in tests and final exams, as well as the respective weightings of concepts and topics for the final grade. And last but not least, the tutor understands exactly which topics and concepts certain exercises and practice questions belong to and where they rank in the complexity tree in terms of "difficulty.'

Through study, practice, and experience, the tutor will have created their own mental model of the complexity tree of eighth-grade math. With this mental model of the curriculum, topic relationships, and an understanding of the complexity tree, the tutor can dive straight into a session, test a student’s existing knowledge with practice exercises and test questions, gain an understanding of the learner’s competence and confidence, and start to adapt the learning journey, in real-time, to the needs of the student.

As required, the tutor will continuously change the pace of the lesson in line with the learner. They will swiftly move to more difficult exercises if the student is already performing well and confidently, or explore deeper—and at a slower pace—concepts the student is finding difficult, utilizing explanations, examples, worked examples, and exercises. And, of course, a good teacher will do all this by making use of spaced-out practice and revision, variation and interleaving—all of the good things we know that “make it stick.” This is what we would consider truly adaptive, fully personalized learning.

If we compare this to the capabilities of the AI contextualization engine, we can immediately imagine how, with the right amount, breadth, and depth of high-quality resources, questions, and assessments, the AI could automatically map out a set or subset of digital learning assets, in line, for example, with a given set of learning objectives or a skills framework. It can do this with a fixed set of resources, already designed for a specific program of learning—or it can pull together the right content from any knowledge hub or learning resources library it has access to, super-fast. In a large organization, this could be an onboarding program for a new employee, which includes basic sales training, basic training on the organization’s customer management system, training on pricing and contract writing, as well as training on a certain range of products and the required product knowledge.

With the contextual piece and the complexity tree in place, an adaptive learning AI engine can take the learner on a journey, just like a personal tutor would. Such an adaptive engine would guide the learner through the complexity tree, from simple to more complex, while the learner learns, practices, and applies their knowledge and skills in order to prove their competence in topic and level of difficulty, respectively, addressing each branch of the complexity tree.

Scalable personalized learning is within reach

Just like a personal tutor, the AI engine can test a learner’s existing knowledge with practice exercises and test questions, gaining an understanding of competence and confidence, and start to adapt the learning journey, in real-time, to the needs of the learner. The engine will also be able to continuously change the pace of the lesson in line with the learner, moving swiftly to more difficult exercises and complexities if the student is already performing well and confidently, or exploring concepts deeper which the student finds difficult. And just like a good teacher, the engine will guide the learner while employing proven techniques such as spaced out practice and revision, variation, and interleaving to improve retention.

What we are looking at here is a fully adaptive, asynchronous digital learning program. Unlike personal tutors, the automated engine is easily scalable and is hyper-personalized to each individual learner and their respective needs and pace of progress.

AI-facilitated personalized learning is a complete game-changer, and it signals the end of one-size-fits-all digital learning, for both standalone formal programs of learning and for any learning based on contextual AI search functions, such as those explored in “Boost Usability of Libraries & Knowledge Hubs with Automation.”

Furthermore, personalization based on AI contextualization empowers and drives peer-to-peer knowledge sharing, as content produced by peers can be utilized easily, without the necessity of tagging or cumbersome organizing. And since the AI contextualizes from the content itself, it no longer even matters what file name a piece is given.

However, the impacts will be even wider: Since the AI adaptive engine will continuously measure the learner's competence and confidence in each topic area and in each branch of the complexity tree, what emerges will provide a real-time image of competence, progress, strengths, and weaknesses, for individuals or for groups, that is easily accessible.

This mapping could form the foundation to drive learning and performance strategy overall, and specifically targeted intervention, coaching, mentoring, training, and workshops, all of which we will explore in more detail in the next article.