For the L&D teams now charged with supporting their organisation's AI transformation, this is unlike anything they've been asked to deliver before.

Meeting the moment requires a major step: L&D has been promising innovation for years across known improvements to the ‘course-first’ model – with limited meaningful success.

In the case of AI, these shifts need to be made for true enablement of the business to work.

It’s time to finally crack three problems L&D has been wrestling with for years:

• The shift from information transfer to behaviour change.
• Achieving meaningful measurement.
• True solutions for continuous learning in the flow of work.

The good news is that this is one of the first cases in recent memory where boards are genuinely invested in L&D getting it right. Research shows that 85% of organisations increased their AI investment in the past twelve months, and 91% plan to increase it again this year.1 That level of sustained commitment creates a rare opening for L&D to solve the problems it hasn't been able to solve on its own - provided it's willing to transform itself in pursuit of the transformation the business wants.

 

AI adoption is exposing the limits of traditional L&D.

The default response for many L&D teams is the one that’s served them for years: a capability gap is identified, a curriculum is scoped, modules are built, and completion rates are measured. That instinct has been working less and less well for years, as organisations demand better ROI and people become increasingly numbed to change in the workplace. And now, with AI, this approach simply will not work at all.

This year, 93% of executives pointed to human issues such as culture and change management as the key challenge to AI adoption, the highest percentage ever recorded. Only 7% blamed the technology.2 IBM's 2026 analysis reaches a similar conclusion: the biggest barriers to realising AI ROI sit in the softer parts of the business, in culture, governance, how work gets done, and how data is managed. AI ambitions, as IBM puts it, "collide with internal realities long before technical limitations."3 All of this research points to the same thing: this isn’t a technical skills issue, but rather an issue around activating AI use as a daily practice.

The traditional L&D model, with its focus on knowledge, content, and course design, has served organisations well in other contexts. When trying to address AI adoption, this model pulls in the wrong direction. Course correcting L&D’s approach to AI adoption and enablement means making critical transformations in three areas, and a focus on the human beings that

 

There are three shifts L&D needs to make for AI adoption to succeed.

1. From information transfer to behaviour change and activation.

For decades, "engagement" has been L&D's proxy for impact, and "modules completed" its proxy for value. AI forces a reckoning with what really matters: are people using it, in their real work, to do things differently than they did last month? And is that translating to value for the business?

Activation is the challenge. It means designing for the first-use moment, the tenth-use moment, the plateau, and the unlearning of old habits. It means building safety nets for experimentation and recognising that none of this work happens inside a course.

 

2. Measurement that the business cares about.

Adoption is what every part of the business is watching, and it’s where L&D can make the biggest difference. And the answer lies less in completion and more in lead indicators of desired behaviour.

Does learning and enablement translate into people using the tools? Are habits changing across teams? Do you have champions leading the charge, and how can you harness this to really nail one of the most human parts of AI adoption? Measuring and reporting on how behaviours are shifting in relation to the take-up of the tech is where L&D can shine, speaking to the people side of the business and giving leaders the evidence that the investment is translating into value.

 

3. True solutions for learning in the flow ow work.

Flow-of-work learning, communities of practice, and continuous learning cultures have been the "best in class" aspiration for years, nice to have if you could manage it. With needs around AI adoption, they become the new floor.

Tools are updating monthly, use cases are emerging daily, and there's no plausible version of this where a one-off training programme keeps people current. The only thing that works is embedding learning into the work itself, into the tools, into the team, and into the rhythm of how people do their jobs.

Describing the shift is easier than making it happen. Fortunately, a reasonably consistent picture is starting to emerge from the organisations furthest along the curve.

 

The habits of L&D teams successfully tackling AI adoption.

Across the organisations we're seeing make real progress, patterns are emerging, and they don't look much like the habits L&D have been relying on.

 

Embed agentic AI in the flow of work.

The best adoption happens when AI support shows up exactly where people are already working, inside their actual workflow, the apps they use, and the tasks they need to complete.

Build communities of practice.

People learn AI from each other faster than from any course. The L&D job becomes convening the right conversations, facilitating exchange between teams, and amplifying what's already working somewhere in the organisation.

Create the space to experiment.

Sandboxes, hackathons, use-case competitions, and protected time to play all matter. Permission to try, fail, and share often matters more than any curriculum.

Content as a launch point, not the be all and end all.

A short video or a prompt library can spark a behaviour, but what happens afterwards is what defines adoption. Treat content as the starting pistol for a longer race and invest in everything that keeps people running.

Stop trying to keep up with content.

You won't manage it, and nobody can. OpenAI pulled its Sora app roughly six months after launching the standalone version, scrapping a billion-dollar Disney deal in the process; any L&D team that had built a course around it was left in the lurch.4 Curating, signposting, and building the organisation's ability to learn from itself is more valuable than maintaining a pristine library that's out of date by next quarter.

Address the numbness first.

Our Generation Numb research found that people are overloaded, disengaged, and exhausted by the constant drumbeat of change.5 If you pile "mandatory AI training" on top of that without addressing the underlying fatigue, you'll hit a wall. Meeting people where they are - emotionally, cognitively, and practically - is the prerequisite for everything else.

 

The AI adoption challenge is also L&D's biggest opportunity.

What makes this moment genuinely different is the alignment at the top. Boards are leaning in, with CFOs asking about AI ROI, CEOs pushing for AI-first strategies, and CHROs under pressure to demonstrate their people function is AI-ready. For L&D, this marks an opportunity to execute on a strategy that every part of the business is connected to.

AI adoption is more than another capability gap to respond to - it's the pressure that finally closes the door on the old L&D model and forces a new one into existence. The moment demands an L&D function that is agile, responds to the needs of the business, and keeps people at the centre of everything it does. But perhaps that’s the operating model L&D has always needed. AI just happens to be the thing that made it non-negotiable.

To learn more about L&D’s pivotal role in AI transformation, visit James Gordon’s talk at Learning Technologies Exhibition and Conference 2026: AI adoption is stalling. Now what? Theatre 6, 30th April, 10:15 – 10:45.

And to learn more about mobilising a numb workforce, check out my talk, Mobilising Generation Numb, at Theatre 1, 29th April, 15:30-16:00. Or download the original whitepaper here.

References:

• Deloitte, AI ROI: The paradox of rising investment and elusive returns, 2025.
• Harvard Business Review, How Executives Are Thinking About AI in 2026, 2026.
• IBM, How to maximize AI ROI in 2026, 2026.
• Reuters, OpenAI drops AI video tool Sora, startling Disney, 2026.
• Sponge, Mobilising Generation Numb, 2025.

 

Josh Cardoz 

Chief Creative and Learning Officer at Sponge