Forum Discussion
Is L&D's role expanding in the age of AI?
We just published a new article exploring how AI is changing the role of L&D.
One of the core ideas is that as more learning happens in the flow of work, L&D's role may expand from creating learning to helping ensure learning is effective wherever it happens.
I'm curious whether that resonates with what you're seeing:
- Has AI changed how people learn day to day?
- Has it changed what your L&D team spends the most time on?
- Where do you see the biggest opportunity, or challenge, for L&D right now?
How is your organization approaching this? I'd love to hear your perspective in the comments!
27 Replies
My answer to your 3 questions first…
Has AI changed how people learn day to day? Yes, but less in the “I did a course” sense and more in the “I asked something and got an answer in ten seconds” sense. It seems formal or informal training is no longer the first place anyone goes.
Has it changed what we spend time on? For me it has shifted the balance from production towards standards and tooling. Less time building each individual thing, more time building the thing that makes each individual thing good by default. That’s the ‘Embed’ step in practice, and it’s the part I’d argue is doing the real work in the article’s model.
Biggest opportunity and challenge are the same item. The article is right that nobody hands you the architect role. But the shift only works if the organisation resources it as a change of function rather than an extra duty bolted onto an already full workload. Otherwise ‘Enable’ just means more content and no more capacity to shape it.
In the article the line that resonated most was the design systems comparison. When prototyping tools put design in more hands, designers didn’t disappear, they built the component libraries that carried their thinking into work they never touched. That’s a genuinely useful model for where L&D goes next.
Thanks for sharing your perspective DShaw. I was hoping to hear from you, given the AI-focused content you've been sharing here lately!
I think you make a great point that this only works if organizations treat it as a change in function, and that probably is both the biggest challenge and the biggest opportunity.
The article talks about making the gap visible to leadership and helping them see the value of L&D taking on this kind of role, but I imagine making that case looks different in every organization.
I'm curious what others here think could help leaders really see the value of that shift.
Thanks Katie-Jordan. On what helps leaders see the value of the shift: in my world it's risk, not capability. I run technical training for a large UK business, and most learning here has always happened on sites or between shifts, long before AI. What's changed is the volume and confidence of informally produced content. I see people generating a "training guide" in minutes and it looks polished, and that polish is the problem. In the compliance heavy areas I work in, a plausible but wrong guide can get someone hurt. That framing gets leadership's attention far faster than any argument about L&D capacity or strategic roles. Once they see the exposure, resourcing the embed step (quality baked into the templates and standards people create with, rather than L&D reviewing everything after the fact) becomes the obvious answer rather than a pitch.
"When prototyping tools put design in more hands, designers didn’t disappear, they built the component libraries that carried their thinking into work they never touched. That’s a genuinely useful model for where L&D goes next."
This resonates with me so much David! I think about this a lot in the community space, too, which I know we were chatting about on another thread. How can I use this technology to get the "best practices" into the hands of others without them having to explicitly learn them, so that the function scales without me being present in every single interaction?
Tagging a few members I'd be interested to hear from: Kate_Golomshtok Thomas_Shayon Frankie KayleneWance Nedim
Here we goooo! AI has been everywhere and it's changing all the time. I feel like my opinions on AI has been on a roller coaster because like anything, there are good parts and bad parts.
- Has AI changed how people learn day to day?
Who doesn't love instant gratification? That's been a big change for learning. I can go to a generative AI model and just ask the question no matter WHAT it is. I need to figure out how to do something in Photoshop? I can just ask AI. What about an Excel formula but I just can't remember what it's called? Ask AI. Before, I'd have to know someone who had those answers or, search the internet.
- Has it changed what your L&D team spends the most time on?
I'd say yes/no. The group I work with does all the pieces ADDIE. We have used generative AI to help with the Design part of the process: outlining, some scripting, a bit of storyboarding but we always have a hand in what it looks like. I can only speak to me but, sometimes I dislike what I get from AI and just go do it myself. But, with that help on the Design portion of the process - it has giving me more time to develop some more effective and fun training on topics that aren't natively fun.
- Where do you see the biggest opportunity, or challenge, for L&D right now?
A big challenge the AI gap that's actually outlined in the article! Now anyone can create training or WBT with the help of AI. That doesn't necessarily means it's good learning. As L&D professionals, we know adult learning Theory, the different processes to create learning, design principals, plain language etc. But someone else who doesn't know those things can ask AI to "create a training on XYZ" get something they see as passable, put it out there not knowing it's not effective. Then we get a influx of bad training others see and think is 'good enough'. It dilutes the good training with sheer volume of 'good enough' training.
Yes, thank you for sharing this KayleneWance! I completely get the AI roller coaster ride.
I love the connection between your first and last points... Anyone can go to AI to get an answer to almost anything, which means they can also use AI to create training. So how do we make sure what they create actually reflects the learning theory, design principles, etc., you mentioned?
That's what makes the idea of the "Capability Architect" so interesting to me. There's this big opportunity to help set those standards and enable others to create effective learning, but figuring out what that looks like in practice is definitely the challenging part.
I also love your point about AI giving you more time to make training more effective and fun. I think we are seeing more and more of that in the community, and it's a great example of how AI can support your expertise.
"It dilutes the good training with sheer volume of 'good enough' trianing"
This is gold KayleneWance!
You hit the nail on the head with your comment "put it out there not knowing it's not effective". The article's model. Enable, Embed, Amplify all happen at or before the point of creation. Nothing in the model checks whether the learning was correct, or is still correct six months later. That's the step I think is missing: a feedback loop. Every analogy the article uses has one. Design systems have deprecation. DevOps has monitoring and rollback. The article borrowed the enablement half of each pattern and left the assurance half out. And its own diagnosis says shelf-life has collapsed. Keep adding content while shelf-life shrinks and you end up with a pile of stale guidance. In my field that's worse than nothing, because people trust it. It was "the training!".
"Here we goooo" is exactly the energy this topic deserves 😂 The AI opinion rollercoaster is real.
The Excel formula example made me laugh because that's such a perfect everyday case. Before, that question either went unanswered or turned into a whole Google rabbit hole. Now it's just answered, and you're back to work in seconds. That instant gratification piece is such an underrated shift.
I really like how you framed the Design phase change too. AI taking the first pass on outlining and storyboarding so you can spend more time making the "not natively fun" topics actually fun is such a good use of the extra time. That's the part that should be exciting, not threatening.
Your last point is the one that's been sitting with me. The dilution problem is real, when anyone can produce something that looks passable, the bar for what people accept as "good" quietly drops. That's exactly why the L&D skillset (adult learning theory, plain language, design principles) matters more now, not less. Knowing how to build something effective is becoming the differentiator, even though it's less visible than it used to be.
A huge amount of learning has always happened outside formal courses, and todayAI has made that visible and measurable for the first time. When someone can ask an AI assistant "how do I do X" and get a contextual answer inside their actual tool, that's a learning event L&D never touched, never tracked, and never designed. It just happened. That's a real expansion of the addressable surface.
Has AI changed how people learn day to day?In short: Yes. AI has shrunk the gap between "I don't know how to do this" and "I did it" to almost nothing. People learn by asking mid-task instead of scheduling training. That's faster, but riskier: getting a quick answer can feel like learning without actually building a durable skill, since the struggle that makes things stick often gets skipped. It's also made learning more reactive, triggered by whatever you're doing right now: which is great for relevance but can crowd out the kind of open browsing where people used to stumble onto skills they didn't know they needed. So the real question for L&D isn't "did AI help people get answers faster": it's "did that moment leave them more capable next time, or just unblocked this time."
Has it changed what your L&D team spends the most time on?
I am not part of a larger department with multiple L&D and ID roles. However, the change I would recommend is spending less time producing content from scratch and more time on review, judgment, and governance. AI now drafts a lot of the first pass: outlines, quiz questions, scripts. So the work would shift from "write it" to "check it's accurate, on-brand, and actually teaches the right thing." At the same time, new work has appeared that didn't used to exist: vetting AI tools, setting data/guardrail policies, and figuring out how to measure whether learning embedded in workflows is actually working, not just whether a course got completed.
Where do you see the biggest opportunity, or challenge, for L&D right now?
The biggest opportunity is that L&D can finally tie learning to real outcomes measuring whether skills got applied, not just whether courses got completed. The biggest challenge is that AI adoption is outpacing the infrastructure and governance to support it, while the skills people need keep changing faster than L&D's operating model was built to handle. So it's less "build more courses" and more "build the systems to know what's working, wherever learning happens."The future opportunity is not just using AI to make more courses. It is building self-improving learning ecosystems that continuously collect signals, analyze them, and adapt in real time. A possible future state looks less like an LMS and more like a Learning Intelligence Platform.
Nedim, this is one of the most thoughtful breakdowns I've read on this thread. Bookmarking it.
"Did that moment leave them more capable next time, or just unblocked this time" might be the sharpest way I've seen anyone frame the actual stakes here. It's such a clean test, and it makes the risk feel concrete instead of theoretical.
The shift you describe, from writing content to reviewing, judging, and governing it, matches what I'm seeing too. AI can draft the first pass fast, but someone still has to catch what's wrong, keep it on brand, and make sure it's actually teaching the right thing. That's real L&D work, it just doesn't look like the work we used to do.
Your closing point landed the hardest for me. Most conversations about AI and L&D stop at "make more courses faster."
You're describing something bigger, a system that's actually watching what's working and adapting, not just a faster content pipeline. Would genuinely love to see what a Learning Intelligence Platform looks like in practice. 🙌
One thing which is quietly funny to me, given this is a thought-leadership piece about L&D bringing quality standards to AI-produced content, try dropping the article itself or some of the replies to this thread into any or multiple LLM and ask it if the content was AI generated.. some interesting results which may be right or wrong but interesting in this context none the less… Perhaps this highlights further the difficulty a casual learner has identifying polished AI generated but wrong content from human authored and fact checked content…. Just my thoughts
My ELH weekly challenge submission, and final course in my AI series, sits very nicely with the article I think!
AI drafts fast, confidently, and about half of it is wallpaper. This course teaches you to edit it, spot the six tells, fix them without swapping one kind of slop for another, and check the numbers a model will invent with a straight face.
Tools used: Canva, Gemini and Rise Canvas
DShaw, wow!
Thanks for the red pen exercise. I learned several things I was oblivious to (now, I'll delay my judgment, when I see a pristinely designed site with the bluish-purplish colors, dotted borders, and stars for the sake of stars).
The interactions you built into the Rise course were phenomenal!
I know you said you used Canva, Gemini, and Rise Canvas. Have you published a "this is how I built it" video? I would love to see the process. Your design gave me inspo for a course I want to build.
I haven't created a 'how I built it' video, but perhaps that's something for another day! The basic workflow is as follows: Create the design and layout of each section in Canva and convert to code also in Canva - Create the interactions in Rise Canvas - Use Gemini to help code the section transitions and combine all the code together into one file (you could use any LLM for this but I use Gemini because its pretty much free) - Upload to Rise as a project zip code block...
This sounds like a very simple process but there are several and often quite complex steps not described (Including the usual ADDIE stuff in addition to all of the CSS complexities that Rise iframes cause)...
What a great way to wrap up the series! "AI drafts fast, confidently, and about half of it is wallpaper" might be the most accurate sentence I've read all week 😂
I love that you're teaching the editing skill directly instead of just warning people to "double check everything."
Naming six specific tells gives people something they can actually act on, and the bit about not swapping one kind of slop for another feels like the part most AI courses skip entirely.
The number checking piece is huge too. I've had an AI model hand me stats with total confidence that turned out to be completely invented, and if you're not already in the habit of verifying, it's an easy trap to fall into.
I already went through this course and it's great, congrats on wrapping up the series! 👏
Hi, Katie-Jordan.
I hope your summer has gone well. Public school has started back in my neck of the woods, which means I get the chance to leave my house even earlier in the mornings to dodge as much traffic as possible driving to the office. 😂😂
Thanks for the tag and the prompt.
Context matters. For example, in my org, my team and I report to the VP of Contact Center. The learning content we create directly supports our 250+ phone agents. Thus, as our business exists today, there is no drastic shift in what we're asked to create (e.g., training on new policies & procedures, product knowledge, and system upgrades, etc.).Regarding this question, "Where do you see the biggest opportunity, or challenge, for L&D right now?"
From my vantage point, I believe L&D is uniquely positioned to help its leaders frame AI use as a corporate imperative rather than "ad-hoc tools spun up in different departments with no real strategic weight behind it."
Companies are burning through cash because their employees are hitting token limits, but no one is steering the AI ship.
I encourage fellow learning leaders who can influence the AI conversation in their org to do so and quickly.
For me, instead of company leaders settling for incremental productivity gains, they should have folks coordinating and executing moonshots. 🚀🌘
"No one is steering the AI ship" is a great callout. There's a big opportunity for L&D to help set standards for how these tools can support more effective training and enablement.
Hi Thomas! Your point hit home.
The "ad hoc tools spun up in different departments with no real strategic weight" line is exactly what I see happening too. People are experimenting on their own, getting mixed results, and nobody's connecting the dots or setting a standard for how to choose the right tool for the right job.
That's actually what pushed me to build Ask Ada. Instead of being another ad hoc solution where everyone just wings it, I wanted something that gives people a clear, reasoned answer for which AI tool fits a given task, so the judgment isn't stuck in one person's head. It's a small example of the kind of steering you're talking about.
On a personal level, I'm using AI to build more engaging learning, faster. I genuinely love using it as a thought partner, it pushes my ideas further and lets me spend more time on the parts of design that actually need a human touch.
I love the moonshot framing too 🚀 Incremental gains are safe, but they're not going to be what separates the orgs that figure this out from the ones that stall. Appreciate you naming that so directly.
One moonshot could be, as Monika Saha calls out in the article, sitting down with company leaders to define exactly what a Capability Architect looks like in one’s organization.
It’s the kind of thing I dream about, for example, using AI to redefine how we do business. If I had my way at my employer, I would create an AI think tank and include people from all levels of the company.
The aim would be something like this:
“With AI, the sky is the limit. Let’s brainstorm (as humans) to reimagine what work gets done, how it gets done, and if we should even keep operating that way in our company. Then, let’s take that output and feed it into Perplexity, Claude, or another generative AI tool. The purpose of giving our brainstorming output to GenAI is to stress-test ideas, refine the strongest ones, and create a plan to begin testing our organizational rebirth as a human-first, AI-augmented entity.”
In an organization like the above, I can definitely see how a Capability Architect (formerly, members of the L&D team), role comes to life.
SMcNicol, thanks for your commentary. I always appreciate your thoughts.
This conversation has really taken off and is so interesting. I'm tagging in a few more community members whose perspectives on this I'd find valuable—
JennMerrill and LauraSpielvogel, since I saw you reshared the article on LinkedIn, I'm curious what about it felt most resonant or what you would add to the discussion.
JeffBatt and DevlinPeck, I'm always curious your take on new technologies since I think you have a really deep engagement with and understanding of AI.
CrystalBass, amalialm and SMcNicol, tagging you in since you've shared some interesting examples about using AI recently.
Feel free to tag anyone else you think would make a great addition to this conversation!Katie-Jordan pointed me to this thread after she left a comment on my Ask Ada post, and it feels like a perfect fit for what's being discussed here 🙌
I built Ask Ada because I kept getting the same question from teammates: "which AI tool should I use for this?" Instead of staying the person everyone pings, I built a tool that reasons through it the way I would.
It doesn't just spit out "use Claude" or "use Copilot," it walks through why, based on the task in front of you.
That feels like exactly the shift the article and this thread are pointing at. L&D's role is moving from just creating content to building the standards and judgment that travel with people into the moments we're not personally there for.
If you're curious, here's the post: Ask Ada
Would love to hear what this group thinks! 😉
