Picture a fairly ordinary training session. The learner shows up, camera on if that's required, answers the poll when prompted, and finishes the module on time. By every visible measure, it went fine. And yet by Monday, none of it seems to have made a difference. Nothing transferred into how they actually work.
This is a kind of disengagement that rarely gets talked about, mostly because it doesn't show up in the usual metrics. Completion rates look healthy. Satisfaction scores come back acceptable. But anyone in that room, including the person running it, could sense that most people were going through the motions rather than actually there.
There's a more accurate word for that than "disengaged." It's compliant. And it's easy to mistake for engagement, because from the outside they look almost the same.
Two things that look identical and aren't
Compliance and genuine engagement share a lot of surface behaviour. In both cases, people show up and respond when asked, and both register the same way in a dashboard. The difference sits underneath, in what's actually driving the behaviour.
Compliance comes from obligation. It's a sense of must, should, or expected to, and it channels cognitive effort into finishing rather than into learning. Engagement runs on something different: a feeling that your presence matters, that your view is genuinely wanted, and that some part of the experience was built with you in mind rather than just delivered at you.
The research is fairly blunt about this. According to a 2024 meta-analysis by Slemp, Field, and Ryan published in PLoS ONE on self-determination theory in the workplace, emotional and cognitive engagement are very unlikely to appear in mandated learning, no matter how well the instructional design has been done. Once a learner feels processed rather than invited, compliance takes over, and real participation doesn't get much of a chance.
Why the industry keeps missing this
At this point it stops being a facilitator's hunch and starts looking like a structural pattern.
Training spend in the US came in close to $100 billion last year, according to Training Magazine's 2025 Training Industry Report, with technology budgets continuing to rise even where overall spend has tightened elsewhere. Despite that, TalentLMS's 2026 L&D Report finds that roughly seven in ten employees admit to multitasking during training, and Gen Z, the largest group of learners entering the workforce, points to staying motivated, rather than access or tools, as their biggest obstacle.
So the money is moving one way and the engagement numbers aren't following it. A large part of the reason is that most of what gets bought is genuinely good at solving the presence problem. Platforms track completion, systems log attendance, and content can be personalised in ways that weren't possible a few years ago. None of that reaches the compliance problem, because compliance was never really about delivery. It's a design failure a level up, in what learners are actually being asked to do and whether that feels like an invitation or an instruction.
The AI question, honestly
AI is currently being positioned as the next answer here: smarter personalisation, adaptive learning paths, real-time sentiment tracking, automated follow-ups. Some of that genuinely helps, particularly around relevance, which is a real driver of engagement in its own right.
But personalisation isn't the same as invitation. A learning path built precisely around your skill gaps is still something happening to you rather than something you're shaping. It addresses relevance well and leaves the compliance question exactly where it was.
So the honest, still-open question is whether AI helps facilitators build experiences that feel like genuine invitations, or whether it mostly helps organisations push out more training, faster, to more people. Those lead to fairly different places, and which one wins out will likely shape whether the next round of L&D investment actually narrows the engagement gap or just makes the compliance experience run a bit more smoothly.
What closing the gap actually requires
If the gap isn't mainly a technology problem, then closing it isn't mainly a technology decision either. It comes down to a handful of choices that no platform makes for you: what you're actually asking someone to do, whether they have any real say in it, and whether the experience gives them a reason to bring their full attention rather than their scheduled presence. Some of that is about invitation. Some of it is about relevance, about pacing, about whether a learner ever gets to make a genuine decision inside the session instead of just following it. None of it shows up on a features list, because it isn't a feature. It's a set of judgement calls a facilitator makes before anyone logs on.
That's the harder truth underneath the investment numbers. The industry can keep buying better delivery. It can't buy its way out of a design problem. Closing the gap will come down to the same thing it always has: the people in the room, deciding what this experience is actually for.
Źródła
Slemp, G. R., Field, J. G., & Ryan, R. M. (2024). "Self-determination theory and workplace outcomes: A meta-analysis." PLoS ONE. pmc.ncbi.nlm.nih.gov
Training Magazine (2025). 2025 Training Industry Report. trainingmag.com
TalentLMS (2026). 2026 L&D Report: The State of Workplace Learning. talentlms.com
Training Industry (2026). Why Gen Z Is Unhappy With Workplace Training, and What L&D Can Do About It. trainingindustry.com
