Artificial Intelligence  

Cutting Through the Noise

There is no doubt that artificial intelligence dominates the conversation right now. New tools, products, and applications are launched almost daily. Research papers, opinion pieces, and headlines flood our feeds. You can hardly open a newspaper without seeing AI mentioned — often with a sense of urgency, excitement, or even fear.

Keeping up is hard. Even if you try.

I consider myself someone who actively follows developments in this space, and still, I struggle to stay fully on top of what is happening. The pace is relentless. The amount of opinions articles and sales pitches on AI is huge. And separating meaningful insights and core developments from poorly informed decisions and short-lived hype has become a challenge in its own right.

So instead of adding more noise, I want to try something different here: step back and focus on what truly matters for L&D.

One thing I want to state very clearly upfront: AI continues to surprise me — in a positive way.

You’ll hear many people say they are disappointed in AI, frustrated by unreliable outputs, hallucinations, or inconsistent quality. I understand that frustration. But I strongly believe the core issue is not the technology itself.

In a future edition of this newsletter, I’ll go much deeper into this, but my position is simple:

AI as a technology will continue to evolve, and as we will continue to create data. AI will continue to improve. And it will continue to amaze us in ways we cannot yet fully predict.

If we look beyond the hype and focus on what is actually happening inside organizations right now, I see two short-term trends that are particularly relevant for Learning & Development: In house large language models and Agentic AI/Process Automation

In house large language models

More and more organizations are deploying their own large language models (called LLM’s).

Public LLM’s like ChatGPT or Claude, are trained on the abundance of data available in the public domain. In house large language models are in essence similar algorithms but this time trained on the data that exists within a company. Think of

• Internal documents like procedures, policies, best practices and marketing

• Internal data available in tools like salesforce, SAP, Workday and your learning tools

• Knowledge repositories like SharePoint

• Even emails and online conversations (naturally with consent of the owners/participants)

The result is in essence all the company’s knowledge captured and made available in a single artificial intelligence. This also allows AI to be embedded into all your internal tools and platforms. Often powering chatbots that support employees in real time by in essence being able to answer any question:

• What is the policy on taking leave?

• What is the procedure on XYZ?

• What are the latest product specification?.

I believe internal LLM’s will create a fundamental shift in how employees learn and obtain performance support.

Instead of employees going to a learning management system, browsing a catalogue, enrolling in a course, and completing a program, they will increasingly:

• ask a question at the moment of need

• receive contextual, job-specific guidance

• get support while working, not before or after working

In many cases, that what we traditionally have labelled as “learning”, “a course” or “a training” will be replaced by AI operating and interacting in the flow of work. This will dramatically reduce the demand for traditional learning and development programs and thus will have a direct impact on L&D budgets and staffing.

According to ChatGPT, around 70% of corporate L&D efforts go into knowledge based programs. Think compliance training, procedural training, but even topics that lend themselves for skill based training like leadership and soft skills are still largely using passive, knowledge based learning formats.

And knowledge based learning programs will in my opinion be the first to be replaced by in house LLM’s. Because why go to the elaborate process of going to your LMS or LXP, search for suitable content, then fast forward to the relevant part, when you can simply ask a question to a chatbot?

So if we consider that 70% of L&D programs are knowledge related, you could argue that

Demand for L&D will drop up to 70%

Agentic AI and AI based Process Automation

The second major trend is the rapid rise of agentic AI — AI systems that don’t just respond to questions and prompts, but actively do things.

Across industries, organizations are building AI agents at an astonishing pace to automate structured, task-driven processes. And L&D is no exception.

Think about what this means for our domain. AI agents can be trained to perform many tasks that are currently run by (operations) L&D teams

• Simple Reporting

• Program Planning and Scheduling

• Program Administration

These are, in essence, structured processes. And they are exactly the kind of processes agentic AI is well suited to automate.

I fully expect that many of these activities will increasingly be handled by AI — initially with human oversight, but over time with far less human intervention than we are used to today.

And yes, I expect content development & instructional design to also be done by AI. In fact, right now, a survey from Synthesia shows “87% of respondents are already using AI” and mainly for content creation.

Now as you know, I’m always looking at the data behind reports like this and as Synthesia is and AI content creation platform, and there were only 421 respondents. So these numbers are not saying much.

However, Donald Taylor and Eglė Vinauskaitė also released a report last year september (2025: The Race for impact) that supports the notion that AI rapidly changing L&D. More importantly, the report explicitly states that AI is eroding the foundations of traditional L&D work, especially content creation:

• AI-generated content is good enough

• It is orders of magnitude faster

• It is far cheaper than traditional approaches

And this is when almost no L&D team has implemented through Agentic AI!

The Implications for L&D Professionals

There is doubt for me that these developments in AI serious implications for L&D and for you as an L&D professional. It’s not a matter if the have, now it’s a matter of what the implications are and how we can start to prepare for what is changing. How we can reposition L&D as an industry and how we can reposition us as a professional.

Because the technical developments are currently faster than ever before, but it will take some time for people and organizations to adapt. And there is where our opportunity sits.

But we have to face reality.

A significant portion of what we currently consider L&D work will either:

• Be replaced by AI-driven, just-in-time support, or

• Be automated through intelligent agents operating behind the scenes

For professionals whose work is heavily centered on operational, administrative, or content-driven activities, this is an uncomfortable reality. And it puts L&D in a tough spot — not because learning becomes less important, but because how learning happens is changing fundamentally.

Next week I’ll share my view on the 2nd major trend that I see shaping the future of L&D: skills — and why the skills conversation now is about much more than upskilling alone, especially in the age of AI.

Tips for Getting Started

Tip 1: Observe where AI is already replacing “learning moments”

I always reflect on how I learn and transpose that to how I think employees work and learn. And I must be honest, I have not taken a formal training in many years. I’m also using a lot of AI to support my work and writing. But I have never done a course on AI. So for me AI is already replacing a lot of learning.

That is why my first tip is to observe for yourself in the next week how and where AI is replacing situations where you normally would

• Search the LMS

• Ask a colleague

• Check a document

• Watch a tutorial

…and instead use an AI tool (ChatGPT, Copilot, internal bot) to get the answer.

Now you experience for yourself: This is the 70% shifting in real time.

You don’t need any permission to do this — but once you see it, you can’t unsee it.

The question then becomes: What learning moments are quietly disappearing — and which new ones are emerging?

Tip 2: Practice “teaching” AI with better prompts, not better tools

What to do
Take one recurring work question you have and improve the prompt, not the AI.

Try now to:

• add context (“I work in L&D in a large organization…”)

• add constraints (“keep it practical, no theory”)

• add one or more examples (“an example from compliance learning”)

Now you experience for yourself: This is how you train AI, even if it doesn’t feel like it yet.

The skill of prompting with intent will matter more than trying every new tool on the market.

Tip 3: Start seeing yourself as a “learning sense-maker” rather than a “learning creator”

What to do
Ask yourself this question at the end of the week:

What part of my value comes from creating content — and what part comes from understanding how people actually learn and work?

No action is required here, just reflection, or an informal conversation with colleagues over coffee or tea.

Why this matters
AI will keep getting better at producing things.
You as L&D professional can step up the value chain from a valuable L&D professional who produces programs to one that can make sense of things.

This mental shift prepares you for everything that follows in the series.