AI Is Not Replacing L&D. It Is Replacing the Parts of L&D That Were Only Content Production.
AI is not coming for learning and development. It is coming for the parts of L&D that were never truly learning strategy in the first place.
That is the uncomfortable truth behind the panic. AI can draft a course outline, write quiz questions, summarize policy documents, generate examples, and turn raw information into something that looks polished enough to launch. If the job is mainly to produce more slides, more modules, more scripts, and more knowledge checks, AI will keep getting faster, cheaper, and harder to ignore.
But that is not the end of L&D. It is a reckoning. The work that was mostly production is becoming easier to automate. The work that requires judgment, diagnosis, design discipline, and business credibility is becoming more valuable.
In my work with corporate learning teams, the strongest L&D contribution has rarely been the content itself. It has been the thinking behind the content: understanding the real business problem, identifying the performance gap, deciding what people need to practise, and designing the conditions that help someone do something differently when the work gets hard.
The uncomfortable part L&D has to admit
The content-production version of L&D has been under pressure for years. Many teams have been asked to respond to business requests by producing something visible: a course, a deck, a toolkit, a checklist, or a webinar. Those assets can be useful, but they can also become a substitute for deeper performance work.
A stakeholder asks for training. The team gathers source material. Someone builds a module. Learners complete it. A completion report goes out. Everyone can point to activity.
The harder question is whether anything important changed.
That question cannot be answered by content volume. It requires a closer look at behaviour. What are people doing now? What do they need to do differently? What gets in the way? What does good performance sound like, look like, and feel like in the actual work?
AI can help with parts of that process, but it does not remove the need for expert diagnosis. It may make weak diagnosis more visible because it can produce content so quickly that teams no longer have the excuse of production time. The bottleneck becomes clarity.
Where AI is genuinely useful
AI can be useful in L&D when it is used to reduce production drag, not replace learning judgment. I would use AI as a support tool, not as the learning strategist.
For example, AI can be useful for:
- summarizing long source documents
- organizing expert input into a clearer working outline
- identifying themes across sales calls, manager feedback, or learner responses
- drafting early scenario ideas for review
- creating alternate examples for different learner groups
- generating learning solutions once the competency, behaviour, and business context are already clear
The important part is sequence. AI should not decide what people need to practise, what good performance looks like, or which behaviours matter most. That still requires human judgment, business context, and learning design expertise.
Used well, AI can reduce some of the manual drafting and organizing work. It should not replace the thinking that makes training relevant, practical, and worth doing.
The risk is not that AI helps with drafting. The risk is that organizations mistake a faster draft for a better learning solution.
The work AI cannot own on its own
AI cannot be fully accountable for whether a learning solution is the right answer to a business problem. It does not own the context. It does not understand the politics of a sales organization, the pressure on a frontline manager, or the gap between what leaders say they want and what the work actually requires.
That is where L&D still earns its value.
The future-facing L&D role is less about producing every asset by hand and more about making better decisions. It means asking sharper questions before the build starts. It means translating broad needs into specific behaviours. It means saying no to training when training is not the right solution. It means helping leaders see that awareness is not the same as ability.
Most importantly, it means designing the conditions for practice. People do not build confidence in a difficult sales, service, or coaching conversation just because they read what good looks like. They need a safe place to try, stumble, adjust, and try again.
Practice is where L&D becomes harder to replace
Practice is one of the clearest ways to move L&D out of the content trap. A good practice experience is not a generic conversation. It is intentionally designed to elicit the exact skill the organization wants to build.
The learner may experience these as short, practical activities. Behind the scenes, they require design discipline. Someone has to decide what the practice is for, what behaviour should show up, what feedback should be given, and what progress should look like over time.
That is not content production. That is performance design.
What I would change now
Leading an L&D function right now, I see the work differently: the goal is not to protect every task from AI. It is to decide what can be accelerated and what must still be owned by learning leaders.
I would also make five practical changes:
1. Stop starting with the asset. Start with the workplace moment that needs to improve.
2. Define the behaviour in observable terms before building content.
3. Use AI for speed, but require human review for accuracy, relevance, and instructional quality.
4. Build practice into any program where judgment, confidence, or conversation skill matters.
5. Measure skill growth and application, not just completion.
The opportunity for L&D leaders
AI creates risk for L&D teams that are seen mainly as content producers. It creates opportunity for L&D teams that are seen as performance partners.
The difference will show up in the questions L&D asks, the solutions it recommends, and the evidence it brings back to the business. The teams that thrive will not be the ones that produce the most material. They will be the ones that help people improve the work that matters.
That is why I do not think the strongest L&D response to AI is defensiveness. It is focus. Let AI reduce the manual load where it can. Then use that space to do the harder work: diagnose better, design better, practise better, and prove the difference.
Beyond Role Plays helps teams move from content-heavy training to focused skill practice. We create the training, scenarios, role plays, scoring, feedback, and rollout support so teams can build the conversations that matter without starting from scratch.