How AI improves the impact of training?

Measuring impact is more crucial than ever for the future of training. Everyone now agrees that the question of “better” has become at least as important, if not more, than the question of “more.” 

In corporate contexts where the challenge is often to train more and better (and usually with the same resources), how do you measure the real effectiveness of your training programs? What are the most common success indicators, and why? How can you improve them to increase ROI for both your company and your learners? And what role can AI play in boosting training efficiency? 

These are pressing questions we all face. Thanks to the contributions of AI and Data Science, new answers are emerging, answers that feed research as well as the curiosity of those most passionate about the future of learning. 

Here, discover how AI enhances training impact through an analysis of one million anonymized learning experiences (e-learning modules and blended programs), carried out by Grégory Vérey, Director of Innovation at Teach Up, and Fabien Marchand, Director of Pedagogy at Very Up.

This is the average reduction in time required to complete and succeed in an Adaptive Learning e-learning module compared with a linear approach. The figure comes from a 2022 study comparing traditional linear modules (created with Rise360 or Storyline) with the same content transformed into Adaptive Learning modules in just a few hours. Reference population: more than 7,000 learners in both cases. 

In a linear module, everyone goes through the same content in the same order, within the same timeframe. In contrast, in an AI-powered e-learning module, algorithms analyze every learner interaction, identify points of interest, highlight high-value content, assess each learner’s expertise level, and adapt the depth of content and exercise difficulty in real time. 

• If the module becomes too difficult, AI restores confidence by offering easier exercises, motivating learners to continue. 
• If content is too easy, AI shortens explanations and streamlines the path, saving time for experts (on average 65% less screen time for the top 20% of subject experts).

With the help of machine learning, it’s possible to guarantee mastery for every learner at a given moment, without traditional final assessments. Micro Adaptive Learning is a game changer in formative evaluation. 

Here’s how it works: as learners go through a module, the technology analyzes their responses and interactions in real time. It adapts the depth of the lesson, the difficulty of exercises (down to choosing the most relevant “wrong” answer to challenge a learner), and the type of feedback provided. 

The result? Each learner reaches the target mastery level defined by the instructional designer, at their own pace and in their own way, supported by personalized feedback after every question.

An Adaptive Learning module optimized with Teach Up can automatically generate up to 14,291 unique learning paths for a single learner, in less than an hour, based on the author’s content and guidelines. 

On average, such a module lasts 14.1 minutes, includes 9 lessons and 14 interactive games. At this “sweet spot,” all three parameters are optimized simultaneously: 

• Mastery: ensuring the training is effective. 
• Completion: motivating learners to finish the module. 
• Time spent: adapting to each learner’s expertise level. 

Longer modules with more content and interactions tend to lower completion rates (learners drop out). Shorter modules, on the other hand, can fail to achieve the mastery target. Adaptive Learning allows both objectives to coexist.

Predictive AI (also known as discriminative AI) is the most widely used but least understood form of artificial intelligence. Based on machine learning models, it focuses on predicting outcomes from input data, hence its name, as it “discriminates to predict.” It can differentiate, sort, and select, in a positive sense. It relies on several core concepts: data, model, training, evaluation, and inference (the process that leads to predictions). 

Soon, AI will also help us: 

• Measure the carbon footprint of training programs before they’re launched. 
• Prioritize learning actions in real time for large reskilling or upskilling projects. 
• Identify subject matter experts at a glance and connect them with those who most need their support. 
• Correlate job performance data with learning data to generate decision-making insights. 

For example, Elevo’s AI automatically matches employee skills with catalog content, recommending the most relevant training courses while still leaving decision power to learning managers. 

But here’s the key: how we present this data, who we show it to, and what we choose to do (or not do) with it will shape the story we write in our organizations around AI and data. Behind every promise of “more” and “better” lie unprecedented ethical challenges.