Al in e-learning: what’s actually working?

Most L&D teams have used AI in some form by now. The question has moved. It’s no longer about whether to adopt it. It’s about whether it’s actually delivering something better, or just delivering something faster.

In our latest webinar, Nelson Sivalingam, CEO at HowNow and author of Learning at Speed, and Talha Faridy, AI Innovation Lead at Easygenerator, joined moderator Ashling Moran for an honest look at where AI in L&D is actually landing. What has worked, what hasn’t, and what L&D professionals should be asking of the tools they invest in.

🎥 Watch the session: Missed it live? Watch the full recording above.

Efficiency is not the same as effectiveness

When you ask how L&D teams are using AI today, the answer is mostly the same: to do what they were already doing, but faster. Content creation, curriculum planning, analyzing evaluation data. The efficiency gains are real.

But Nelson drew a line between efficiency and effectiveness that came up throughout the whole conversation.

“There’s a big difference between doing a thing fast and doing the right thing. And right now what we’re doing is doing a thing fast, but we’re not necessarily evaluating whether it’s the right thing.”

The risk is not just wasted time. It’s that AI makes it easier to scale the wrong things. If your content wasn’t landing before, producing more of it faster does not solve the problem. As Nelson put it:

“What we might actually see ourselves doing is scaling the very problem we are hoping to solve. One of the challenges is there being a ton of content. And when you’ve got a ton of content, discovery is one of the biggest challenges. By being able to create a lot of content that’s not great quality very fast, what you’re actually doing is scaling the problem you already had much, much more quickly.”

There’s also a harder-to-recover-from consequence. When learners engage with poor-quality AI-generated content and don’t find it useful, they form an opinion. Getting them back after that is difficult.

Talha added that unrealistic expectations are part of what drives this. Many authors assume they can hand a task to AI with minimal context and get something production-ready back.

“A lot of authors have wrongful expectations of what AI can do. They feel that they can just delegate without a lot of context and a lot of groundwork, and then expect an output that is completely ready for production. Whereas oftentimes that’s not the case.”

What a real measure of value looks like

If speed isn’t the measure, what is? Nelson’s answer goes back to what L&D is actually for.

“L&D’s outcome isn’t to create content. It’s a means to an end. Our job in this profession is to help people do their jobs better and help them grow. It’s as simple as that.”

That means the question isn’t how many courses were shipped or how many hours of learning were completed. It’s whether the skills that matter to the business actually got built.

Nelson framed this through the lens of what he called the talent crunch: organizations struggling to hire specialized skills that are in high demand and short supply. The only sustainable response is to upskill people internally. That makes skills-building the output that matters, and it’s the output that most L&D measurement still doesn’t track directly.

“The question is not about how many hours of learning someone did or what content was completed. It really comes back down to: did we build the skills we needed in order to achieve those objectives? That’s really the measure of value.”

Context is what AI is missing

One of the clearest practical points from the session was about what actually makes AI output useful. Generic AI tools are reasonably good at generic information. What they’re not good at, by default, is company context.

Talha was direct about this:

“Large language models are really good at generic information. But one thing they are still not great at by default is company context. Your company context, your business goals, the outcomes you’re looking for. That is the kind of context you want to bring into the AI. That prep work needs to be done prior to working with AI on any tool.”

This applies to course creation too. The ease of building something with AI also makes it easier to build the wrong thing.

“It’s easier to build things now, but it’s also easier to build the wrong things. That concept applies to course creation. You’re building something to drive meaningful outcomes. It’s easier to build training, for sure, but it’s also easier to build the wrong training for the wrong people in the wrong formats.”

The prep work that prevents this is defining learning objectives upfront, being specific about the behavior change you want learners to have, and giving the AI a clear picture of what good output looks like before you start. In Easygenerator, this is built into the Course Guidelines feature, which lets authors give EasyAI explicit instructions about goals, content preferences, and source material before it generates anything.

Author-first AI vs AI-first content generation

This brought the conversation to a distinction that ran through the whole session: the difference between AI that starts from the author’s intent and AI that starts from content generation.

Talha described how Easygenerator approaches this, using Bloom’s taxonomy to anchor learning objectives before any content gets built.

“We ask the subject-matter expert or the instructional designer what exactly they want their learners to do after they finish the course. What is that behavior change in terms of application that they want to drive? Once that is defined upfront, along with the knowledge the learner needs, that steers the AI in the right direction and overall uplifts the quality of the content.”

Nelson added a useful frame for what this unlocks. In the old model, an L&D team would go to a subject-matter expert (SME), collect expertise, take it away, apply instructional design thinking, and come back weeks later with a course. AI-assisted authoring changes that timeline.

“As an SME, I can go to Easygenerator and work with the AI to say: I’m dumping my expertise here, but you’ve got the expertise of the pedagogy, and the best way to frame and structure this. It would have taken you weeks to get to something that was pedagogically sound. But now we can get from ‘here’s expertise’ to ‘pedagogically sound learning resource’ very quickly.”

That said, neither speaker suggested this removes the need for human judgment. The power users of AI are the ones who iterate, who treat the first output as a starting point, and who bring their own context and expertise into the process at every step.

L&D’s new job is to engineer the context, not the content

If SMEs can now create content directly with AI support, what does that mean for L&D’s role? Both Nelson and Talha pointed in the same direction.

Nelson’s framing was memorable: in this world, L&D teams are effectively managers of AI. And a manager’s job is to set the bar and create the conditions for others to hit it.

“It’s not necessarily L&D’s job to engineer the content. It’s L&D’s job to engineer the context so everyone else can create more useful resources with very little friction.”

That means defining what good looks like, setting course guidelines that any SME can follow, choosing the right pedagogical frameworks, and building governance structures that keep quality consistent across the organization without burdening every SME with instructional design decisions.

Talha made the same point from the product side:

“The L&D team can set that framework and steer things in the right direction. If you don’t, it’s just another situation where SMEs, who are already stretched, now also have to learn how to create quality learning. AI can be super helpful here, where L&D teams define what good looks like, and then SMEs can use that to create content without it adding burden on them.”

The recognition piece also matters. Ashling mentioned something she sees work in practice: putting people’s names on courses, calling out strong examples, and giving public credit to SMEs who do it well. It sets the bar visibly and gives others something to aim for.

Whether AI is built in or bolted on matters more than you might think

A question from the audience during the session touched on something that came up throughout: how do you tell whether AI is genuinely built into a tool or just added on top of an existing product?

Nelson’s take was that the answer usually shows up in how the AI performs across the full workflow.

“When you’ve got AI native, you’re essentially applying intelligence at every step in that process. So you get the efficiency gains, but also the gain in effectiveness by applying that intelligence at every step. As an add-on, it’s really difficult. You’re trying to polish the last mile with AI, which is what makes it very difficult.”

Content creation, as Nelson pointed out, involves multiple steps that benefit from intelligence at each one: curriculum design, structuring individual pieces of content, building in scaffolding, applying pedagogical frameworks. A bolt-on solution can help at one or two of these points. A platform where AI is native can support all of them.

The practical test he suggested is simple: how effective was the AI at helping you do the specific task you were trying to do? Not which AI sounds most impressive, but which one actually helped you solve the problem in front of you.

Talha addressed the pricing dimension of this directly. Some vendors have responded to the cost of AI by passing costs on through higher prices. Easygenerator’s position is that AI should be included in the base price for everyone.

“We strongly believe that the future for most authors and L&D people is AI-native. Our position is that everyone should be empowered with that technology. Hence, for our core platform AI, we don’t charge extra for it. It is included in the base price. That is the core reason for it, to empower everyone and work with them towards that future of being AI-native.”

Where this goes next

The final chapter of the conversation looked at the next 12 months. Talha pointed to agentic AI as the direction of travel: AI that moves through a workflow iteratively, using different tools and capabilities to accomplish a task, rather than producing a single output in response to a single prompt. More effective, more context-aware, better at getting to something useful without requiring the user to do multiple rounds of manual correction.

Nelson’s outlook was bigger in scope. He described a gap in L&D that has never been closed: the connection between learning and performance. Most organizations can’t tell you, when performance drops, which skills are missing or why. And even when they can identify a skills gap, they often don’t know which learning will address it.

“With autonomous agents that can go through that loop all by themselves, you can now look at work signals, infer what the performance gaps are, figure out what skills are missing, connect you with the right relevant learning resource, then monitor the work data to see whether it actually changed the performance. All in one autonomous loop.”

He called this the vision of a self-improving company. L&D sets the guardrails, and the system handles the loop.

The bottom line

AI in L&D is delivering real efficiency gains. The organizations seeing the most value are the ones treating context as a first step, keeping humans in the process rather than removing them, and measuring outcomes rather than output.

The tools matter too. AI that’s built into the authoring workflow from the start behaves differently from AI added on top of a legacy product. Understanding that difference is part of evaluating whether a platform will actually help you or just help you produce more content faster.

👏 Huge thanks to Nelson Sivalingam  for a genuinely honest and practical conversation.

Speakers webinar transcript

Ashling: Let’s go ahead and get started. Thank you all for joining. If you’re coming back, we really appreciate it. And if this is your first time, welcome in. Today we’re going to be discussing AI and e-learning and what’s actually working. My name is Ashling and I have the pleasure of hosting this session, and I’m joined by Nelson and Talha, who will introduce themselves in just a few seconds. This is a follow up to a call that we had in October 2025, and that was all about AI and e-learning. And this time we are starting from a slightly different place. Instead of looking at should we be using AI, we’re going to be focusing on is this actually working? Without further ado, I’m going to hand over to our guests today. Nelson, you’re first on my screen. So would you like to introduce yourself?

Nelson: Sure. I’m Nelson, one of the founders and CEO at HowNow. At HowNow we’re a learning platform that helps companies figure out what skills they have, what skills they don’t have, and connect people with relevant learning so they can build the skills they need. I’m also the author of Learning at Speed. Be great to know if anyone’s read it on this session, if you haven’t, it’s a book that gives you a practical guide and frameworks on how to accelerate the speed at which you learn in your organization. Great to be here. I’m really looking forward to the discussion.

Ashling: Well, thanks for joining us. I’m excited to see what you can add. If anyone has read the book, pop it in the chat, I’d love to know. Talha, would you like to introduce yourself next?

Talha: Thanks, Ashling. So Talha here, leading the AI product efforts at Easygenerator. So I’m responsible for the AI vision, product strategy, and working with L&D teams and subject matter experts to achieve the dream of Employee-generated Learning, and how AI can be applied in authoring, e-learning, but also beyond. And we’re exploring that space here at Easygenerator quite rigorously, I shall say.

Ashling: Exciting time, a lot is changing, so I’m excited to have some time today to pick your brain. Before we dive in with our first question, I’d love to hear from all of you joining us today. In one word, I’d love to know how you would describe where your team is with AI right now. I’m going to pop a poll in the chat. You can choose which option you think resonates the most. If there’s something that we haven’t listed, feel free to also type. Okay, with the early answers, it seems like many teams are still experimenting, or even a step forward, and scaling. So yeah, a lot of teams are currently experimenting. It’ll be interesting to see what you’re experimenting with, what’s working well. So today what we see is most L&D teams have actually started to use AI in some form, and the question has moved. It’s no longer about whether you should be adopting AI, it’s about whether it’s actually making learning better or just making it faster. So we’re going to get into this honestly, what has worked, what hasn’t worked, what the quality difference really looks like in practice, and what L&D professionals should be asking of the tools that they invest in. So, to kick us off, Nelson, I’d love to know your insights on how L&D teams actually use AI today, and if it’s working.

Nelson: Yeah, I guess looking at how people are using it today, I think it’s largely for efficiency gains. What’s clear from, actually, there’s a great study that Donald Taylor does looking at AI usage in L&D, which clearly shows that we’re essentially using AI to do the things we were already doing, but to do them faster. And so that’s specifically around content creation, or helping us with planning a curriculum, or mapping out content. And then comes down to analyzing data. So if you’ve got large sets of evaluation data or learning engagement data, being able to analyze that and make sense out of it are probably the two primary areas. But the challenge is we’re trying to be more efficient with AI, but what we’re not seeing is uses where it’s making L&D teams more effective. And that’s really where the challenge is right now. To your question around, is it actually working? There’s a big difference between doing a thing fast and doing the right thing. And right now what we’re doing is doing a thing fast, but we’re not necessarily evaluating whether it’s the right thing or not. And that comes back to a lot more than just applying AI to some of the tasks that we were doing already. There’s questions around how it aligns with the overall business strategy. Are you measuring the right things to know whether you’re having the desired impact on your organization or not? And so I think that’s where we’re currently seeing the challenges. I saw almost 10% of the poll said they’re skeptical, and what I’m seeing is a lot of skepticism comes from you’re not immediately seeing the productivity or uplifting outcome. And we’re jumping to the conclusion that that’s the fault of AI or the AI tools, rather than asking the more fundamental questions of, are we applying it right, and is there a more fundamental strategic problem to see if it’s actually working or not. So there’s definitely efficiency gains, but have we seen a significant uplift in effectiveness? We’re yet to see that.

Ashling: Yeah, but that would, I think, makes sense. A lot of people who’ve answered in the poll have said that they’re in that experimenting phase, so trying to test the water, see what’s working and see what’s not. And hopefully that’s what we’ll take a closer look at today. Talha, have you noticed the same, that we’re doing things perhaps faster but still have to unlock that efficiency piece?

Talha: Yeah, absolutely. I think from my perspective, how I’m seeing things, there’s two main categories as to where the need of AI and the current situation is falling into. There is one where there is a limitation in resources when it comes to L&D teams, right. So they have these large backlogs, a lot of tasks that they have to accomplish, a lot of courses to shell out. Large volume of courses, so one problem. And the second one is the quality, right. Where, of course, in order to overcome this bottleneck in resources, one of the ways is to delegate content creation to subject matter experts. That being Employee-generated Learning. And there, of course, there’s a question of quality. How do you maintain quality of output, instructional design grade, learning that is being created across the board. And I think that’s where L&D teams are looking towards AI to address these problems, and more so towards the production side of things. Content production, that’s probably the biggest one. And there I second with Nelson, for sure. And I absolutely agree, we see this a lot, that producing things faster doesn’t necessarily mean that you are doing it the right way. And also leads to false expectations and false outcomes sometimes. A lot of authors have wrongful expectations of what AI can do. They feel that they can just delegate it without a lot of context and a lot of groundwork, and then expect an output that is completely ready for production. Whereas oftentimes that’s not the case. So there of course needs to be strong governance, strong strategy, alignment with business outcomes, for AI to really be effectively deployed throughout the content creation lifecycle. So yeah, that’s what I’m seeing on my front.

Nelson: And just to add to that, right, I think it’s an important point that no one gets to keep their job for being efficient at not delivering value. And so that is the risk right now, around whether you’re actually delivering an outcome that the business wants, or are you just shipping a lot of stuff? And what we might actually see ourselves doing is scaling the very problem that we are hoping to solve. One of the challenges is there being a ton of content, and when you’ve got a ton of content, discovery is one of the biggest challenges. By being able to create a lot of content that’s not great quality very fast, what you’re actually doing is scaling the problem you already had much, much more quickly, and therefore making it harder for yourself. And also it creates a negative feedback loop, right. Where if people start engaging with your AI generated content and it’s of low quality, it’s very difficult to get them back in, because they’ve tried it, they didn’t find it useful, they didn’t find it relevant, and now they’ve formed that opinion around this content. So the next time you’re launching a course and saying, come back, I’ve got more stuff, they’re less likely to engage because the last time it created a negative feedback loop. And so, as much as you might be drawn to the efficiency of churning out content faster, actually it can have some damaging impacts down the line, where you could make it harder for yourself to engage people in learning moving forward, if you don’t ensure that it’s of a high quality and relevance.

Ashling: Yeah, that makes sense. So making sure that you have the quality there, it’s something that learners want to engage with. If they take a piece of training, they see the value for themselves and they want to engage again. What I find interesting there, Nelson, is if speed is not the same as value, what should L&D teams use as a real test of whether AI is working or not?

Nelson: Yeah, I think it goes back to the holy grail of what L&D is about, right. L&D’s outcome isn’t to create content, it’s a means to an end. And our job in this profession is to help people do their jobs better and help them grow. It’s as simple as that. And so then you start breaking that down to go, what are we looking at here to know whether we’re delivering that value to the organization or not? Now, if you look at the macro environment context, right now we have a massive talent crunch, which might not be that obvious given a lot of the news headlines right now are talking about massive layoffs at organizations. And therefore it’s easy to think that unemployment rates are going up. But actually, if you look across, like the UK, for example, in the US, unemployment rates are not going up, they’re going down. And what we actually have is a hiring drought for very specialized skills. A great statistic I saw was a recent survey found that for technical workers, about 70% of technical workers had multiple job offers at the time where they accepted a job. Now, what that tells us is for very specialized skills, there’s very high demand. In fact, there is a higher demand for certain skills than there is supply, which means it’s very competitive and expensive for organizations to hire the skills they need to get the job done. Which means the only sustainable way for companies to do this in the current environment is to be able to upskill and reskill their own people. Which means that’s the goal, right. The goal is, how do we ensure in our organization we are building the skills that we need to be able to get the job done? Which means now we need to be able to understand what skills we don’t have. And that becomes our priority. These are the objectives that the business wants to achieve, and in order to achieve those objectives, here are the capabilities that we need to build in the organization. And whether we’re delivering value or not comes down to, are we building those core capabilities or not? And so the question is not about how many hours of learning someone did, or what content was completed. It really comes back down to, did we build the skills we needed in order to achieve those OKRs or objectives? And that’s really the measure of value. It’s less around speed to create content, but it’s speed to create content that was relevant for building the skills that we needed to deliver value to the organization.

Ashling: Makes sense. So if somebody is joining today, would your advice be, if you’re at the beginning of this process, find out what skills you need and then that would inform the rest? And if you already know what skills you need, start measuring whether your training efforts are helping meet that skills gap, or see value being delivered because these skills are met?

Nelson: Yeah. I’d almost look at it like a pyramid, right. And at the top of the pyramid, if you imagine everything the business does is at the top of the pyramid. Every business has a mission, and every business has a strategy on how to achieve that mission. And then every business has a set of KPIs or objectives or OKRs to see whether they’re on track for delivering that strategy or not. Let’s say that’s the top of the pyramid. At the base of that pyramid, from an L&D perspective, you have learning initiatives, right. And typically what you see in organizations, there’s a gap between what L&D are doing and what the business is trying to do. Typically we try to connect the dots to the top of the pyramid retrospectively. I’ve pushed out a bunch of content and now it’s like, did it deliver value? Well, let me figure out how it did. And we try to retrospectively connect those dots, and that becomes, it’s a bit like playing bingo. Someone’s called out a number, some people have got the matching number, they win. But the reality is most people don’t have a matching number, so they’re losing. So every time someone loses, that’s a waste of time, resources, and money. When you play L&D bingo, what you need is a more certain way of being able to connect those dots. And the missing gap in that pyramid is what are the skills that the individuals and the teams need to have to achieve those KPIs, right. So the process I would go through is, up front, you speak to the business about, okay, what are the OKRs or objectives here that really matter most to us? Let’s pick this one, let’s say it’s increasing customer retention. Then we break that down to go, what are the job roles that are going to most influence these objectives? And what skills do they need to have in order to be able to deliver the tasks that are going to drive that objective? Once we know what skills they need to have, now we need to figure out what skills do they actually have. Do they have those skills? Let’s say they need to have problem solving skills at a level three. Do they have that or not? And if they don’t have those skills now, it becomes around, those are the focus areas for you to work on. So what content do you need to create to improve problem solving in your organization, or what resources might you already have that you just need to make it more accessible or engaging for the right people, to be able to make the most out of those? And now that you know what the baseline of that is, you now track that performance objective metric to go, are we moving towards that or not? Is customer retention getting better? If it’s not, then we need to change track. We need to look at, is it the content, is it the discovery of the content? But the key metric we’re looking at is that, and the reality is in too many organizations, we’re not even measuring that performance metric until way too late. And the reality is you achieve what you measure. If you think about it, up until five years ago, who was conscious of getting 10,000 steps? No one really thought about whether I did my 10,000 steps or not. But now that we can measure whether we do our steps or not, now we all care about doing 10,000 steps. So you achieve what you measure. And to start with, you need to start measuring the impact of learning on skill development, and skill development and performance in order to achieve it. If you’re not measuring it, you’re not going to achieve it.

Ashling: Makes a lot of sense. And I also think that perhaps, Talha, this links back into something that you mentioned, that some teams here, some organizations are already measuring this, or at least they’re on the journey towards this final destination. And now they have a skill, they need training for the skill. Maybe there’s only one or two people in the organization that can currently fulfill a certain need or a role. So they start to create training, but they’re expecting AI to do too much, or they’re expecting too much out of AI. So Talha, my question would be, what kind of context actually makes the biggest difference when you are leveraging AI to help with the learning piece?

Talha: Yeah, that’s a very good question. I think also reading through the chat, the participants in the webinar, I think there’s one thing that’s very clear, is that large language models, though they’re getting better at it, they’re really good at generic information. So high level information on most of the topics, they’re pretty good at some of the technical things too. But one thing they are still not great at by default, that’s very important, by default they’re not great at, is company context. Your company context, your business goals, the outcomes that you’re looking for. That is the kind of context that you want to bring into the AI. And I think that prep work needs to be done prior to working with AI on any tool that you’re working with. So once you’ve identified the skills that you want to target, and coming up with the right learning objectives, keeping the outcome in mind before you actually go start building the e-learning, so having all of that documented upfront and using that to guide your creation efforts, is something that really, really matters. I think also one really good example to give here is that, even if you look at it from a product perspective, it’s actually easier to build things now, but it’s also easier to build the wrong things. And that concept also applies to course creation. Because you’re building something to drive meaningful outcomes. And it’s easier to build a training, for sure, but it’s also easier to build the wrong training for the wrong people in the wrong formats. So that prep work, of course, bringing your company context, outcomes that you’re looking for, and drafting the right learning objectives, putting the right context into the AI, that really, really matters.

Ashling: I would say, yeah, makes sense. I think perhaps there’s a step between what Nelson, you were just explaining, about the gap and preparing and knowing what we’re measuring, and then also building the content. And I know that there’s probably people joining this webinar today who are in that step, that in between step. So perhaps, Nelson, you could help us identify how to move forward. So how would you suggest that, once you identify the skills and the business needs, and you need to then decide what learning intervention will make the biggest difference, what would you advise L&D teams to do in that situation?

Nelson: Yeah, so part of it is done in understanding what are those skills gaps. What are the skills that you need now? The skills then break down into what are the tasks and behaviors that they should be able to do as a result of having the skill. Now, a lot of that comes from problem discovery. I like to use the example of using a product mindset to this, because I think a lot of it does apply. And a big part of product building is problem discovery. How much time do you actually spend trying to understand what the actual problem is? Why do we have a performance problem? What tasks and behaviors are these individuals not capable of doing, or need to get better at doing? And that’s where working with the SME really, really matters, to be able to understand that context of how this skill will be applied. Now, a great example of this is if you take a skill like communication, for example. How communication will be applied is very different for a salesperson within their context to how it will be applied for an engineer within their context. So being able to understand how that skill will manifest itself as a task or behavior will then help you ensure that you’ve got the right resources to be able to do that. Now, when you come to resources, again, it goes back to that question of, what is the right format? It’s not the same format for every type of skill, and that’s typically what we get caught up in. But what we need to really think about here is an ecosystem of resources that are right for different contexts, different roles, and different skills. It’s not about having one course that’s going to solve the problem for everyone who has that particular skill. And that’s where I think you lean more in on curating, and essentially even bringing different SMEs to create content. Leveraging, say, Easygenerator, to empower them to create that content is a form of curating, where L&D’s role is shifting from being almost a content factory that churns out all of the content, to essentially empowering subject matter experts to feel empowered to create the resources that you need to plug those gaps. So that’s what that bridge looks like for me. I think one of the critical steps is, before we rush into creating content, is really doing that upfront problem discovery to understand what is actually missing.

Ashling: It makes sense. So yeah, there’s a lot of things happening, and it’s interesting to see how it’s evolving over time, and how doing that upfront discovery can really help, and having a plan going into things can make a difference. I know that this seems to be resonating, there’s a lot of comments in the chat. I just seen one come in from Jodi, and I think it’s interesting about how Jodi has used AI to help create content every day. So for doing things like summaries, outlines, scripts from screen recorded videos, or structuring curricula, based on the output, then you can go in and validate and iterate with better prompts. So here, using AI to help with the quality piece, and I know that that’s a concern for many on the call. Talha, it’s something that you’ve mentioned as well, so I am curious if this is something that’s really resonating with anybody, what would you suggest that they do to evaluate whether an AI feature is actually improving the learning quality, or if it’s just saving time?

Talha: I think one thing is for sure, I would second, we’re seeing this pattern a lot, in terms of generating helpful summaries, outlines, and then of course getting that first output and then iterating on it, not expecting the AI to be the silver bullet. Let’s say, for example, you give it a prompt and boom, in one shot you have your course ready, that does not happen. The reality is that you have to go and iterate with the AI, and a lot of the times the missing context that was not there initially, that is what is brought in in that phase, and then you perfect the content and make it publish ready. And that’s something we see a lot. Actually the power users of AI, the ones that use it most effectively, are the ones that do follow this kind of workflow. Whereas the ones that kind of expect the AI to do unrealistic things probably drop off after the first use. The ones that continue in the iteration cycle are the ones that get the most out of AI, that’s definitely for sure. And in terms of building effective content, I think also it’s important, going back to Nelson’s point, also having that end behavior change that you want the learner to have basically in mind from the get go, and designing your content around that. Within Easygenerator, we use Bloom’s taxonomy to help with this process, where during the content creation cycle we basically use Bloom’s taxonomy to help with the learning objective creation, and that then inspires the rest of the content that is created with AI, in terms of, we ask upfront from the subject matter expert or the instructional designer that’s using our tool to create learning, as to what is it exactly that you want your learners to do after they finish the course? What is that behavior change in terms of application that you want them to drive? And once that is defined upfront, along with the knowledge they need to know in order to be able to execute on that action, that then steers the AI in the right direction, and overall uplifts the quality of your content. And also helps with picking the right formats, because there’s a lot of different formats you can consume learning in. So again, going back to those goals that you set up front, really helps in helping AI determine what formats to pick. So, for example, for actionable scenarios, maybe role plays are a very good format, whereas for just understanding and remembering, maybe text blocks are the more preferred method. So that is the kind of configuration that you have to do up front, in order to steer AI in the right direction, and of course, create more content that is more effective.

Ashling: Really interesting. So having a human in the driving seat and AI, I see that it’s resonating. Somebody liked it to cooking versus a microwave, or using AI as a sidekick. So, yeah, if a person is involved, that’s where people are already seeing the best outcomes. Nelson, I’m really interested in your point of view. Let’s say teams are investigating this, they want to get people, humans, involved in the learning process. What role should SMEs play in helping L&D understand what the actual performance problem is, so that they can best prepare, as Talha had mentioned?

Nelson: Yeah, so I think what the old model was, is as an L&D, let’s say instructional designer, or as an L&D manager or professional, I would go to an SME to understand, give me the content, give me the expertise, right. So the SME would kind of brain dump all of the expertise, and I, and L&D, would take that away, and I would tinker and apply pedagogy and come up with a course around that. And what AI, combined with authoring unlocks, is that bottleneck of a very lean L&D team having to be responsible for doing that. Right now what you can do is, as an SME, I can go to Easygenerator and work with the AI to be able to say, I’m dumping my expertise here, but you’ve got the expertise of the pedagogy, and the best way to frame this and structure this and do that. So what you’re now getting the benefit of, is it would have taken you weeks to get to something that was pedagogically sound. But now we can get from here’s expertise to pedagogically sound learning resource, very quickly right now. That said, then the question is, what is L&D’s role? And I think it’s a very fundamental and important role there, where L&D sets the guidelines, the framework for what does good look like. And that’s not necessarily, I have to build it. And a good kind of comparison here is, I often say, when I think about what a manager’s role is, a manager’s role for me is someone who sets the bar, and then someone who creates the conditions for you to hit the bar. That’s the only two things a manager needs to do. And I think in this world, essentially we’re all managers of AI. And for me, L&D is a manager of AI that’s specifically designed for L&D. And therefore it’s your job to set the bar for AI, and create the conditions for it to hit the bar. And the conditions to hit the bar here are things like context, that was described before. What have you done to make it easy to be able to bring in that context? Because you can’t expect the SME to bring in the context every time. Are there things like, how you might think about this, almost like a skill MD type file? What are the kind of instructions or files that are universal for the organization, in terms of how we see the world, what we like, what we don’t like? How do you bring in that context, so the SME doesn’t have to do that all the time? Are there certain pedagogical frameworks that you lean towards, that you want to make sure that AI is aware of? So that context building, it’s a big task, and you only realize how important it is when you try creating stuff without that context and see how terrible it is. And that’s really the role of L&D, it’s not necessarily to engineer the content, it’s to engineer the context, so everyone else can create more useful resources with very little friction.

Ashling: I really liked how you put it, setting the bar and then creating the conditions so that you can reach or hit the bar. Talha, given your role in shaping EasyAI at Easygenerator, I would love to know how those lessons have influenced you along the way, on a journey from author-first AI versus regular AI content generation.

Talha: Yeah, I think, in terms of the history with EasyAI, from its inception to what it is today, it also resonates in general with how AI itself and the perception around it has, with time, changed. Of course, when it first came, people had expectations that you would just give it a prompt, and with one single prompt, you would be able to get your output. And then there was like, okay, can you give it a few examples of what are great outputs, and then you would get a better response. Then came the idea of projects, for example, you can then put things in projects and then it will manage your context. Then we came to the era of Claude Code, we have now MD files getting stored as context, and then we have skills nowadays. So we see a clear pattern, things are moving towards, again, going back to Nelson’s point, steering the AI in the right direction, setting that bar, and having skills, for example, having multiple bars that you set for different areas or domains or different things that the AI should do, and then having it follow those. And that’s exactly what author-first authoring offers, where you steer the AI in that right direction, you set the conditions for it. For example, within our EasyAI suite, we have something called Course Guidelines. And within the Course Guidelines, we allow the author to give instructions to the AI as to, what are the content preferences, what are the goals, what are the source files? And it’s basically one of the jobs of the AI to actually read into those guidelines and always refer back to them, whenever it’s making a decision on what to create. And that is extremely important, because then that helps AI understand what is good for me, what is the good output going to be like, and then follow those instructions. So I think this is the common pattern we’re going towards now. And I think from an L&D perspective, again, seconding Nelson’s point, it’s all on the governance side of things. Setting those standards for the AI to then effectively operate across the board. Any SME can pick up the content creation, and then you have this governance layer which is helping experts use AI in the best way possible. Otherwise, then it’s just again building the wrong things at a very fast pace, without any direction and quality.

Ashling: Yeah, I think you hit on something that’s definitely resonating with me personally, and I’d be curious if anybody else feels this too, but AI is becoming such a buzzword, and all these tools have AI now, that it’s really important that we start looking at what AI can actually do, how it’s built into a tool, or is it tacked onto a tool? So perhaps we can also widen up this conversation. Nelson, given the work you’re doing at HowNow, I’d love to get your perspective here too. When you’re evaluating AI learning tools, what signals to you that this is something that’s genuinely built into the product and it’s going to add value, rather than something that was just tacked on afterwards?

Nelson: Yeah, I think it comes back to the same fundamental question of, this is not about being AI first, right, you’re not using something because it’s AI. And I don’t think with any technology that’s the right way to look at it. It fundamentally comes down to what’s the problem you’re trying to solve. And I think the famous line goes, you love the problem, not the solution. And I think that’s key here. Otherwise you can be massively overwhelmed with technology for technology’s sake. But once you’ve aligned on what is the problem you’re trying to solve, it then comes down to what is the tool or the AI tool that’s best placed to be able to help you address that. Now, where it matters, whether it’s AI native versus add on, is with an add on, for a certain level of task, it’s very limited, because there’s a very legacy approach, and then you’re trying to polish the last mile with AI, which is what makes it very difficult. But when you’ve got AI native, you’re essentially applying intelligence at every step in that process. So you get the efficiency gains, but also the gain in effectiveness by applying that intelligence at every step. That’s not to completely rule out bolt on solutions, I think in some scenarios, depending on what you’re trying to achieve, a bolt on might be completely fine for what you’re trying to do. But actually when you look at something like content creation, the reason why it matters is, although we call it content creation, actually there’s multiple steps happening in content creation, from understanding what is the appropriate curriculum design for this, to what are the contents of each thing within this, and how do you build in dynamic scaffolding into this, how do you bring in different pedagogical frameworks into this. There’s multiple things that are happening to do this. And so as an add on, it’s really difficult to do that, we really need to build in AI from the foundations of how you go about creating content. I think the way to evaluate it is, how effective was it at helping you do the task that you were trying to do, rather than trying to look at it as, is this AI better than that AI. Which AI was best at helping you solve the problem that you had in the first place, is the way I would look at it. It really comes back to that mapping, knowing what your goal is at the end, and perhaps not getting distracted by something that’s new or shiny, but whether it’s actually delivering you value, whether it’s helping you move towards your goals.

Ashling: I do see a question from Victoria that came through the chat which I think ties in nicely here. Victoria was explaining how SMEs are now using a lot of AI, they’re creating things like documents, and they’re thinking, hey, this is really great, this is really useful. However, it’s not always actually that helpful, especially if we take pedagogy into account. So, Talha, love your insights here. How can we start to educate subject matter experts on the process, and the need for at least review, when it comes to learning or creating learning content?

Talha: I think more so, I wouldn’t say it’s probably the right approach in terms of teaching the subject matter experts how to create high quality content. I would say it’s the L&D, I think the L&D team can set that framework and steer the right direction. Again, the concept of the Course Guidelines, setting the bar, having some skills, and then having subject matter experts use them. Otherwise it’s just another one of those issues where, let’s say, for example, the subject matter experts do not have the time and the resources to create this content, it’s just going to be another one of those elements that now you have to teach them and have to learn how to create quality learning. And hence I think where AI can really, really help is where L&D teams really take it and have that governance layer, and define what good looks like, and then SMEs can use that to create content, and essentially then it doesn’t add burden on them. So in issues of Employee-generated Learning at scale, SMEs haven’t got the time to take the burden of creating learning. So instead of adding another burden of them having to learn how to create good learning, you create that governance layer, to help them read and leverage it. And that’s where AI can be super helpful.

Nelson: And just to add to that, I think if you went in saying we need to have a review from a learning perspective because we just do, that’s probably not going to land well. And I think you do have the onus on you to explain why, and to back that with data, because otherwise it just seems like a box that needs to be ticked for the sake of ticking that box. And so you need to capture the data to go, where are we, why is it not useful, or is some parts of it useful? And you’re saying, look at this data, it’s not landing here, or it’s not doing X, Y, Z in the way you want it to, that’s because of this. And so I think kind of enforcing that is just a way for people to feel like it could come across discouraging, and you don’t want that to be the message that lands, versus you want to encourage people who are putting in the effort to do this in the first place, which is great. It means they genuinely see learning as a solution to the problem, which is a great place to start, because quite often people don’t even get to that point, they don’t realize learning is the solution to the problem. And so it’s really being tactful in going, here’s the quantitative plus qualitative data that shows that this learning is missing this, and I can help you with that, this is my expertise, let me support you on this, this is what you could have done, and use it more as a coaching opportunity rather than an I told you so opportunity.

Ashling: I like that, so yeah, a coaching opportunity. And more and more we see L&D, as things are evolving, working in these support roles, that their function now within the business is supporting learning and making that possible, and making sure that learning connects to value.

Nelson: If this is something that you’re trying, from working with different organizations that are on this journey, something I see working really well is supplying that support, but then also having some recognition. So it could be putting people’s names on courses, and saying, hey, Talha, you created a really fantastic course on AI, he followed our checklist, and now he’s got rave reviews, or having a competition, or just recognizing and referencing learning. So those who are doing it right are then encouraged to keep doing that, and those who maybe are just turning out documents that aren’t so useful have something to aim for. So again, setting the bar and then helping them reach it.

Ashling: I see another question in the chat, we’re talking about AI and tools, how do we support AI, but there is something interesting as well, it’s about how some tools are tacking on AI and then also additionally adding on some extra pricing, or they’re not absorbing that pricing, and then there’s a question of, is this useful, is this something that we need? Talha, I’d love your input on that before we wrap things up for today, on the pricing, absorbing the price, how do you approach this when you are looking at different tooling?

Talha: Well, it depends from tool to tool, when it comes to offering. I’ll give our position right at Easygenerator, we strongly believe that of course the future for most authors and L&D people is AI native. It is a reality that AI is basically embedding itself into key workflows, regardless of your domain, in most domains it has had a very significant impact. And our position is that everyone should be empowered with that technology, and we of course want to be on that journey with our customers, with the L&Ds, with the subject matter experts, and learn how AI is working for them, with them. And hence, from our position, for our core platform AI, we don’t charge extra for it. It is included in the base price. And that is the core reason for it, it is to empower everyone, and to work with them towards that future of being AI native, and learn with them on how that looks like in the future. So that is our position, and it’s not possible for every tool to do that, it depends on where they’re coming from in terms of how AI native they were, or how ready they were to take on the AI revolution. Because some tools of course have had to pass on that cost of AI onto the customer, whereas we have absorbed it and brought customers along the journey, and we’re of course on the other side.

Ashling: Yeah, I guess it goes back to how AI is being used in the tool, is it something that’s adding value, is it tacked on, will it change your experience? So that’s some really helpful insight. I saw we have some time left, so of course we always want to look to the future, and then we will leave some time for questions as well, so if you do have questions, make sure to drop it into the chat. Talha, perhaps you can kick us off when we look at the future, where do you see AI going in the next year or so?

Talha: Yeah, lots of different things definitely changed with time, of course that AI has gotten great at different domains in different areas. So it all started with just working with text, then it became really good at images and videos, and now it’s really good at many different kinds of tasks and workloads, and working with different multimedia. I think definitely the future for now, which is quite evident, in the direction most of the tools out there are taking, as well as at Easygenerator, the direction that we’re taking is the direction of agentic AI. And it taps into a lot of the problems we’ve been discussing today, the problem of not providing enough context, quality being one of them, getting better output upfront, those kind of different things. And when I say agentic, it’s again the iterative nature of AI, AI having the tools, and it creatively going through these different tools, leveraging them, accomplishing tasks, and working with the author or the user for their particular task and workflow, and accomplishing that. And in that whole journey it then leverages different parts of the context that need to be leveraged, different skills and different tools to accomplish what it needs to accomplish. So I think that’s a more effective way of applying AI, rather than one shot. And I think the industry is realizing that.

Ashling: We will move to the questions, but in the last minute or two, Nelson, do you agree, is there anything else you see in the next six to 12 months when it comes to AI and L&D?

Nelson: Yes, I think one of the things we’ve never really been able to do in L&D is be able to connect the dots between learning and performance. If you look at most organizations, when performance drops, most organizations can’t tell you what skills are missing and what was the reason for the performance dropping. Let’s say by some miracle you do know what skills are missing, you don’t often know what learning is appropriate for that skill level. And to be able to connect those dots, and let’s say learning does happen, very rarely can you say whether it’s working or not. So this gap between learning and performance we’ve never been able to close. And the reason for that is, in order to be able to close that gap, you first need to be able to see work, you actually need to see people work, or work data, to be able to see whether they’re doing something well or not. And then you need to have the intelligence to be able to look at that work data and break it down to go, based on what I’m seeing this person do, they’re good at this, but they’re not so good at this. And then once it figures that out, it needs to be able to give you the exact right help to be able to get better at that task that I wasn’t able to do. And now this typically has been very fragmented, you needed different humans and different systems, and there was a massive coordination cost between all of this. Now, with agents, but more importantly autonomous agents that can go through that loop all by itself, you can now look at work signals, infer what the performance gaps are, figure out what skills are missing, then be able to connect you with the right relevant learning resource, then monitor the work data to see whether actually it changed the performance or behavior, in one autonomous loop. This is what we’ve right now got with organizations that we’re working with using HowNow, and this is where we see it go, is really that performance improvement loop will be completely autonomous, with guardrails set by L&D, but you won’t have to do that legwork, which is incredible, because if you can autonomously improve performance at speed and scale, it really opens up this vision of a self improving company.

Ashling: Which I think is quite exciting, I think so too. And I think you’ve led us nicely into some of the questions that are coming through the chat, and perhaps you can help, and Talha, if you have a point of view on this one, you can jump in. Sanjo has asked, how do you distract from short term revenue generation strategy in your L&D organization without qualitative goals or targets, not to follow shiny objects and AI, but to define a decentralized or creative strategy first? So yeah, how do you perhaps separate the trees from the woods, defining a strategy and not just hopping on the latest trend, any thoughts?

Talha: It’s a very good question. And I think, yeah, one thing that’s definitely coming out of that is that there’s of course a problem at a higher level as well, which is the lack of understanding of how AI functions in general, and the tangible benefits it can bring, and where its limitations are. That kind of leads to those unrealistic expectations, and then prioritization of short term gains, rather than defining a strategy first, I think that shift needs to happen. I think it needs to come from, it should be bottoms up, of course, within an organization where you would have power users of AI, because most organizations would have these power users of AI that are really, really embracing the technology. And I think it’s kind of, maybe they should take shared responsibility to drive that knowledge up there and become subject matter experts within their companies, to really help the leadership realize where AI works, where AI doesn’t, and then work their way from there, and then prioritize a creative strategy. I think that would be my take, curious from Nelson’s angle what that would look like.

Nelson: Yeah, I mean, I’ll build on that, going back to actually tying it in with another question that came in around getting buy in, I think they kind of overlap. You’re not, the tool and technology is secondary, right, the primary thing is, this is the company strategy, what do we need to be able to deliver that company strategy, what are the capabilities that we need to build to do that, that’s how you align your strategy and get the buy in. I typically recommend, and a lot of the organizations we work with, this framework, the three whys, for building your business case. And the first why is why now. And the why now is, most of us, if not all of us, hate change. We hate change. And the reason why we hate change is because change carries a lot of risk, personal risk, social risk, financial risk, there’s a lot of risk, that’s the reason why we hate change. But there’s only one thing that gets us to change when we hate change, and that’s a really big change, right. When a really big change happens, it makes the current status quo unacceptable. And when it makes the current status quo unacceptable, we therefore have to change in order to seize the opportunity, or to navigate away from the threat. And I think the first part of getting the buy in is explaining what is that big change that’s making the current status quo unacceptable, and that’s your why now. Now, once you’ve done your why now, you get to your why, your solution why, the thing that you’re proposing. And that’s where you talk about, okay, I get your why now, but what are the obstacles right now that are stopping you from addressing this problem? You’ve got to name the obstacles before you jump to the solution, otherwise you’re just sending a solution, but you’re leaving it to me to connect those dots. Now, once you’ve named those obstacles, now you talk about your solution, that solution could be HowNow, Easygenerator, AI this, AI that. But when you’re talking about that solution, you talk about it in the context of, what are the obstacles that this solution is going to help me overcome? Now let’s assume you’ve done that, then you come to your last part, which is the why not. The why not is essentially, no business case or proposal is solid until you’ve talked about all the things that can go wrong and you address that. And one of the most common things that come up right now is the fear of messing up. That phobia is the most common reason why people don’t do anything about something. And so you need to address, why do we have this fear of messing up? Is it the lack of expertise, is it the lack of bandwidth and resources? And you build that into your business case in order to get the buy in, and align it to that strategy. So that’s typically the framework we work with organizations to be able to put forward whatever it is that the solution you’re trying to get buy in for.

Ashling: Yeah, and I would say working with clients as well, on the customer value side, when they’re building business cases, makes a lot of sense, so building that out, and I hope that helps. I think you’ve also helped answer a question from Callum about creating buy in, so Callum, three whys, you heard it here from Nelson. Before we wrap up, I also see a question from Monique. Monique is using AI to create some e-learning, and noticed that after regenerating a few different times you can get different results, and is wondering if there’s any guidelines on settings to create the best quality content. Talha, with your expertise, perhaps you could provide some insight here.

Talha: Yeah, I think we’ll try to share some general guidelines, I think of course there are very tool specific ones, but then there are some general things we can do. I think it’s important to understand that AI models by nature are not deterministic, and it’s absolutely the case that in many of your iterations with AI, or workflows that you’re going through, you probably would come up with an output, and then the second time you probably come up with a completely different output. And there are ways to, of course, try to get more deterministic outputs. Of course starting with what you’re working with in terms of prompts, writing very better and detailed prompts, making sure that your prompting techniques are, you’re not giving big prompts, for example things like do this or do that, but rather giving more details as to what exactly you’re wanting to achieve from the output. And if you want something consistent, then making sure your prompts are also consistent, following some sort of a template, and then of course your source content. If you’re working with a RAG based solution where you’re uploading documents and other contexts, making sure that is not too big, and not too over detailed as well. Making sure that your documents and your source content is cleaned up, well structured, and ideal for ingestion. Not something that is like a three or four hundred page book, or a two page document, and expecting it to create a nine page e-learning. So you need to have a good understanding of what you’re working with, in terms of the tool’s capabilities, but then also aligning your source document with that. Prompting is probably one of the key skills here, and then of course cleaning up your source content and making sure it’s AI ready, those are the two things. And from a learning perspective, also going to an example from Easygenerator as well, we have a feature called Course Builder, where we have a section where the learner is defining their goals. So again, there also we really encourage the author to be very specific with their goals, like what exactly do they want the learner to do, or want the learner to know, what topics, what are the outcomes? And being detailed there really helps AI steer the content creation in that direction, and pick up the right things from the source document.

Ashling: Monique, I can pick this up offline as well to see if it is something that was more Easygenerator specific, I can also share some more specific guidelines with you as well, but I’ll make sure to reach out. I think that’s bringing us to the end of today, I noticed we’re nearly at the hour, so before we wrap up, any last words, Talha, or any last words, Nelson?

Nelson: I can kick it off. So it was good to see experimenting as one of the most voted things in your poll, and I think that’s the most important thing right now, I’d say everyone needs to be the chief tinkering officer in their organizations. And what I’d recommend is to tinker in public. It’s not just about you experimenting and tinkering with tools, but I think what’s also important is to build the culture in the organization, to share what you’ve been tinkering with, what you learned, what worked, what didn’t work, and to do that in public. And I think that’s the best way to drive the culture of AI experimentation and adoption in your organization, there’s no better way than tinkering in public.

Talha: Yeah, you summarize it nicely. Yeah, that’s definitely something that I would really encourage, I think that’s really helpful for the organization, and then shared learning of AI is where really the progress will happen. From my side, also very, very excited as to where the industry is headed, there’s a lot of exciting stuff that’s happening, every other day there’s a new AI innovation that happens and it takes the whole social media by storm, and actually quite interesting things that are happening. So what it means for L&D teams and subject matter experts, we will find out in the upcoming months.

Ashling: Let’s see, in the next six to 12 months we’ll be here again reviewing what comes next. But for today, thank you so much to Nelson and Talha for a really honest and practical conversation, I enjoyed it a lot and I have a lot of ideas that I want to go implement now. And I also want to thank everyone who joined us and sent in questions.

Transcript produced from VEED subtitles. Speaker attribution assigned based on context and role.