474 — So long 2025, what's hot for 2026?

Hello and welcome to the MindTools L&D podcast that we can show about work, performance and learning. I'm Ross Garner and this week we are reflecting on 2025 and looking forward to 2026 with our friend Nelson Shivalingam from HowNow. HowNow are a MindTools partner.

If you're a HANL customer, you can access our content via their platform. And it's always a delight to have Nelson on the show. Hello, Nelson. Hey, Ross. Thanks for having me back on this show. Always a pleasure. And we're joined by my friend and colleague, Ross Dickey. Hello, RD. Hello Ross, hello Nelson. So let's start with a look back at the major trends of 2020-25 and of course...

AI remains king. It has maintained its position on top of Don Taylor's global sentiment survey, the annual survey of what L&D leaders think is hot in L&D. And a key aspect of the kind of AI debate is anxiety.

So I'm curious what you both have seen this year in terms of anxiety around AI. I think anxiety is pretty... pretty fair right we i talk about with friends who are other tech founders in the space and i believe we're in the space and sometimes it's overwhelming for us right and we're you're you're hearing about things firsthand versus if this is not your day job and you're not necessarily

necessarily developing AI products. I can imagine why there would be a bit of anxiety. But I think the first thing is just to accept that... you don't need to and it's probably unlikely you're going to stay on top of every ai at launch that's coming out i mean like one week you'll have

Google announced about 30 things. And then the next thing you've got someone else saying, we've got those 30 things, but better. And you don't really need to know all of those things. I think an important thing I take to anyone who tells me a bit anxious about this is the only thing we can really do is tinker. right just just be the chief tinker officer in your organization and whenever you get a chance play around with things and get comfortable with it don't shut yourself away i think that

The negative outcome of this anxiety is, I've also heard some people saying they're just so anxious about the rate of change, they're just going to shut themselves away from it. That's like, no, no, no, that's counterintuitive and you're probably going to end up feeling mine.

uh anxious about it um and so yeah just just tinker um with it is what i'd say but yeah it's acceptable it's okay to feel anxious it's okay yeah i think it's natural to feel anxious i think when you look on uh the likes of LinkedIn and there's so much, there's just so much content around AI and I think it's easy to feel like everybody else is doing something more sophisticated than you are and that you're falling behind.

And you'll never possibly catch up because you just found out about this new model and it's already been superseded by something else. And there's this other new tool. And I think what I have found partly from reading Don's reports. which we're going to come on and talk about is that I think most people are doing fairly similar things. I think to Nelson's point, don't bury your head in the sand.

or think that it's not going to amount to anything. I think I've been very conscious of not trying to get dragged too far into the all-out AI skeptic or all-out AI booster. pack which tends to be one or the other i think on on social media um but yeah just start tinkering with things experimenting a bit but also don't worry too much that everyone else is doing cooler

more impact driving things than you are in L&D because they're probably not. Everyone's trying to figure this out I think at the same time. You're quite weird, you're Ross as well, because while everyone else is creating AI products, you created a human product, you had a baby. And I think we just heard him in the background there making his first appearance on the show. Yeah, Killian's first appearance. I actually didn't hear him. Hopefully it's not too loud. I just tuned that out now.

So I actually, a lot of our clients have been asking what others are doing because there is this sort of perception of like, you know, everyone's doing better than I am. And so a few months ago, I pulled together a kind of...

as a kind of like a sort of media pack kind of like summary of like things that are going on. So I thought I'd just like sort of share some of the stuff that I find and that I've been sharing with clients. So Thompson Reuters reported back in June that only 25% of organizations had a visible AI strategy.

So if you don't have one, you're actually in the majority. And so you don't need to worry too much about that. TechRadar reported in October that three in five workers say they use unapproved AI tools at work. I was shocked. Certainly not something I would do. That'd be hard. Yeah, I don't think we are.

In September, there was an HBR article that went viral reporting that work slop was making work more difficult. And so work slop was this sort of passable looking AI garbage that is created and then shared. But it actually then requires someone to parse that garbage and work out what they're meant to do with it. It actually slows people down. And this was tied to this piece of work from MIT Media Labs.

95% of organizations say no measurable return on their AI investments. And then on the flip side, there's this kind of weird stories about people who are... pretending to use AI at work because they're worried about not being seen as cutting edge, but they're actually not doing it. They're doing it manually, but saying it's AI because they want to be seen.

There's a lot of noise. There's a lot of anxiety from different angles. But on the more positive side, there was another survey from Don Taylor and Egley Van Eskaita. More than 50% of respondents in the L&D field said that they were actively using AI in their work. And this is the first year, the service number three years, is the first year that that number has crossed the 50% threshold. So I thought maybe if we share some practical examples of how...

HowNow and MindTools are using AI, would be an interesting place to start. So, Nelson, do you want to take that first? Yeah, I've got to say, I was actually quite surprised that it was only 50%, to be honest, because I think if you're... a knowledge worker let's call it in that general terms I've been very surprised if you're not working with some form of general purpose AI assistant open net on the side, even if it is. So I don't know whether it is more of a disconnect in terms of.

people are using it and it has come into their kind of day-to-day work, but they're not seeing it as a standalone piece in itself. Because I find that number quite surprisingly low for Iran. Yeah, yeah.

even based on the sample size of conversations I had with L&D teams on a regular basis, they're all pretty much in some shape or form using... your kind of track gpt's or gemini's of the world or co-pilot has come into the world i think in the study the so it was actively using as distinct from experimenting

It's a group that we're experimenting, but then over 50% were actively using it, meaning it's a thing that they use in their day-to-day workflow rather than their, they're not tinkering. You use the word tinkering. They've gone beyond tinkering. They're now using these tools deliberately. And we've seen some, I was all about Pro for HowNow, Pro for the terms of products and how our customers are using it, but also internally. But externally, how we're seeing from a HowNow perspective is...

using it for things like skills mapping. So that's been a really, really big one. We've seen a wave of adoption of skills in terms of how do I map my workforce to skills? How do I map my learning ecosystem, the resources I have to skill?

And this almost, you know, one CHRO is telling me this has been a long time coming. And they're glad they waited as long as they did because they really thought it was going to be like a 10 to 12 month project to be able to get to where they've got to now, which is...

the entire organization speaking a kind of universal skills language. So that's been a big area where AI is at net. And data analysis, right? I think that's been one of the incredible areas for us that we have this AI analyst tool, which just means... not just your L&E team, but you've got managers now asking questions around.

Does my team have the skills they need? You know, are they actually learning things related to their role or not? And they're able to ask that in plain English, get the answer back and analyze the data without having to do. 10 exports and pivot in an Excel to be able to get the answer. So I think that's been great. And then we've got our guru, which is the AI work coach, and that's really kind of unblocked.

and become a primary interface for how people are trying to acquire knowledge and practice and get coaching and feedback. So I think that's been incredible to watch. And I still think we're very, very early in terms of what that adoption engagement is going to look for. And internally, I would very much describe ourselves as kind of early adopters of this tech, right? The nature of being in the space we are. Some of the great things we've done is around essentially agents for cross team.

collaborative workflows, right? Like for example, a sales handover to customer success or collating. feedback from all the customer calls that we're having and then being able to prioritize and align that to our roadmap and, you know, being able to flag any risks or particular areas. And we use agents for identifying those gaps and providing those insights to us.

And the same thing with our kind of commercial process, being able to identify when a particular opportunity is getting held up. So it really has made it easy for us to not only get insights, but also act on it quite quickly. And then from an engineering perspective, we pretty much use it end-to-end for our product development process. So we do everything from AI prototyping, which I think has personally been a game changer, of being able to get a feel for a product experience before.

goes into actual build. And then AI coding, we're using three different tools for different purposes, but that's really accelerated how quickly we can develop and put it out there.

That's probably the core areas. I think one of the challenges, and I'm sure many can relate to this, is just getting the time. You can probably see a tool and go, ooh, we could do A, B, and C. But if you haven't got... the slack in the organization or the time carved out for people to work on these problems and leverage AI for it.

It doesn't matter how great your AI strategy is and whether you've made the tool readily available to everyone and they've not got the bandwidth and it's not been prioritized, then it's hard to actually apply it. Yeah. There's a temptation, we're coming to this later on I think, but to use the word AI interchangeably with magic.

And it's not magic. There actually is effort involved in embedding these tools. And of course, there's not just the, now you have AI embedded in your workflow, but then there's also the, well, that changes everyone's workflow. And getting people to change the way that they behave and change their habits and norms and have a shared understanding of how things should work, that bit's difficult as well. 100%. I mean, that is, this is like a huge...

change management channels. That's what it is, right? Yeah, globally. It's a huge global change management channel. In some ways, that's what learning and development is about, right? And so that's why I do think this is prime time.

for kind of L&D and people teams. And it's unsurprising that there's a big demand and ask on L&D and people teams to be able to get involved in supporting this change management. Because I think where I've seen it go... not so great is is you know it's kind of it led but it led is very much around you've all got licenses now and you can all access it on your machine go figure that's not change management

With a security layer, there's normally this sort of like anxiety about like sort of security. Yeah, of course. You mentioned AI agents, which I think is something that we've seen a lot of buzz around this year. I think there's varying definitions of what actually qualifies as an agent. Without getting too much into the weeds, would you be able to talk just a little bit about...

Well, maybe for the benefit of the listeners, defining how you understand what an agent is. Don't play that card, Ross. It's for your benefit. It's also for my benefit. I can tell you, our interpretation of this is something that can run. essentially a multi-step process or workflow where it also involves making decisions, right? It's using the knowledge and insight you've got.

to autonomously make a decision for what is required in that next step, right? So it's not a process workflow as in, here's the five steps and just go do those five steps. It requires someone to make a decision. learned this from this customer call and therefore I've inferred this from the analysis of that data which now means the optimal thing to do is this thing and this thing and that thing, right? And that would typically require a human to look at that insight, analyze it.

come up with a set of decisions and the next steps. But what you get in an agentic workflow is that end-to-end, if it's a 10-step workflow, every step where that decision is made, AI is able to make that given the access to right context. and knowledge, right? And that is a game changer in so many different ways because otherwise you're bottlenecked by bringing the human into the loop in something that required

such a small amount of time, but it was still required a human to move it along to this next step. And if you kind of compound that for every step in that workflow, then apply that to every workflow across the organization, you can... see why it's such a productivity driver to be able to have an agent take on that responsibility. Hopefully I did a good job there, Russ, of kind of explaining our interpretation of it. Yeah, no, super helpful. I think that's, it's the...

The kind of thing that would be ideally suited to those kind of administrative tasks, which in... That you hit. In Don and... Yeah, well, I mean, we had Heidi Kirby on the show earlier this year, and that was kind of her perspective on how you should leverage AI from an L&D perspective is like, what part of the process do you...

least enjoy? And how can you bring AI into that? I mean, administrative tasks behind content production and learning design tasks was the kind of top use case in Don and Agla's data. I think it's a whole, I want AI to... do my laundry and empty the dishwasher, not create my art piece. I think what I find, we always see this one, there's a new breakthrough of technology. We go for the low-hanging fruit, right? Because our imagination...

doesn't know yet how far it can go. So it's limited by the fact that... let's do this thing, but using AI. But actually, even beyond the admin, I'll give you an example of how we use agents within our products and our customers are using it, where an agent can look at your calendar. see that you've got a one-to-one coming up with your diet report. It has concepts from your talent data that you're a new manager.

And then prior to that one-to-one happening, it finds a window in your calendar. It sends you a nudge to say, do you want to quickly practice that conversation before you actually have the real conversation? You practice it. It gives you feedback. You go do the one-to-one. after the one-to-one and now does you again to go do you want to quickly reflect on how that went now that is what i would describe a multi-step workflow and

And only an agent that is making intelligent decisions around that could deliver that end-to-end experience. And so it goes far beyond automate an existing process to essentially self-generate learning.

That is timely and relevant in a way that just wasn't possible before. Love it. I'm going to move us on from AI because there's more than one trend in the past year. It doesn't always feel like that, but there has been. I think one of the most surprising things that we've seen this year, I'm curious what you've found.

Nelson, in your conversation with clients, is an increase in face-to-face and instructor-led training, which is kind of the opposite to the AI approach. It's like, oh, what do we bring out? a human to train these people. There was a report from Hansley Fraser earlier this year that showed that in-person training was a top delivery method among survey response there in 2025. My IntuSkini have never done more instructional lead training than this year. We actually...

We've created a manager skills workshop. So it's 12 workshops aligned to our skills framework that long-term listeners won't know that we have. And we invited various clients and prospects to come and try it.

one of these workshops we had to end that campaign in 24 hours because we had 100 people sign up good problem to have because someone actually has to run this workshop there is a limit to how and this is the difficulty where the structure led it's difficult to scale can't get an agent to do that yeah

Exactly. But there's a real update for it. As I'm curious for what you've seen, Nelson. Yeah, I think that... is also unsurprisingly i think it aligns with the macro trend we're seeing right there's um a lot of data that shows there's high demand for in-person experiences in general right like people are going to more concerts and and more festivals and more experiences and i think

that micro trend is probably tripling down here. And I think what you get is there's certain things you get from this experience that you're not getting from AI. And I think some of those are very social. a social status associated with being somewhere where other people are not. right and and so being out in this train session training session with a bunch of other people great opportunity for a selfie i can share it there's promo there's a whole bunch of um

added social elements and experiences that you just don't get from the digital experience right now. And I think because it's become they are more rare versus the digital experience that we have. I think there is a tendency to go towards the thing that you don't get as much of, right? There is the scarcity that's in play right now. So I think that's probably driving it. I think it would be dangerous to come to the conclusion that...

There is an increase in this delivery method because it is the more effective way of delivering learning. I think that's a dangerous conclusion to jump to straight away. But I think there's definitely... some things you get out of this experience that you... can't get from digital and like the sense of belonging if i've done it with my team and i'm here um you just can't get that speaking to ai even if it's a group chat with ai you're not going to be able to get that in the same way um when so

I think there's a lot to it. And I'm a bit believing that with AI, one of the things it's done is it's reduced that tolerance for anything that is generic. And one size fits all and not instantaneous and too much of an asynchronous piece, right? Which means things like two-hour courses that are asynchronous and one size fits all. Our tolerance is at the lowest it's ever been because AI has got us used to super personalized, instantaneous responses. The only other thing that can meet that demand.

other than AI, it's humans. When I have a one-to-one interaction with a human, humans are pretty good at personalizing the response instantly. to my context and who I am in that one tongue ratio. And so I think it's unsurprising that we're going to lean more on humans. So I think just as much as AI is going to be instrumental in L&B...

communities and bringing groups of people together and that human-to-human interaction is going to be a big part in meeting that demand. I think the thing that's going to fall into the valley is probably courses still have a place. they probably have a place where you need a structured path to go from zero to one or to get to a certain milestone. But beyond that, experiences like this will play a part.

Yeah, and I think that the remote and hybrid piece is kind of layered on top of that. So that there's the, not only is face-to-face instructor-led training, it's an opportunity to build that sense of belonging. It's an opportunity that exists less. commonly in the day-to-day world of work? Yeah, I think the other way to look at the Hemsley-Fraser data, I can't remember from that report whether it was an increase or whether it was an increase in the previous year.

Right. So I think that does reflect kind of macro trends as Elson was saying, but I think it also comes back to the first question we asked about this sort of fear of missing out. I think a lot of L&D teams are... are still delivering training face-to-face. I think it's easy, I think, if you listen to industry podcasts or read industry newsletters or attend conferences to think that...

Everything is shifting to some sort of AI-powered experience, but a lot of L&D teams are still delivering programs face-to-face or virtually in a classroom setting.

I think that's the other way to look at that data point. And then another key trend is market consolidation. So Workday acquired Sana Labs. Learning Pool acquired the authoring tool Elisadot. Here at MindTools, we acquired Kineo. We're going to talk about that. all year. MindTales Kineo is our new company name. It took us a long time to come up with that, but we thought MindTales Kineo was probably the simplest way to explain it.

Nelson, why do you think this consolidation is happening and what does it mean for L&D leaders who are looking to work with suppliers? The first thing I want to say is it's great to see so much growth and success coming out of Europe. And so I'm all for it. And congrats to everyone who is involved in all of those deals and M&A activity happening. It's good to see it. I mean, it's a combination of things, right? So you've got your big incumbent companies who are in the market.

And they have a couple of problems. One is if they're a tech company, they've got a legacy tech stack, which might be harder to build and drive an AI native roadmap on because of their legacy tech stack. but they need more product to sell to their existing customer base. Because across the board, what we're seeing is companies are generally getting smaller. You're probably not getting the license growth that's tied to seats as you were getting before.

Less growth from license growth. So you need more product to sell to existing customers. How do you get more product? If you can't build it fast enough, you buy. And so you're buying more product to sell to existing customer base. And on the flip side, I think there's, you've got this.

the innovative disruptive startup scale-ups and they're able to build on new frameworks much much faster and execute against the roadmap but the thing they don't have that incumbents have is the distribution right like they've already got a huge

customer base to be able to sell these products to you. And so it's kind of a good marriage, right? Like they've got great products, they need customers, these guys have got customers, not enough great products, they come together, you now distribute the product. Now that's the ideal case, and I think that's what we're seeing.

This is a cycle you often see in the market anyway. Bundling, unbundling happens often in the tech landscape. Now, that said, I do think L&D leaders... and generally anyone buying software needs to be very very wary. of this it makes sense for the incumbents and the scalaks here but does it make sense for l&d leaders now we've been here before where you get this one big company who does one product line really well

And then they throw in cheap versions of bundled other products. Here's an LMS for pretty much free because it's not our core products. And that's where you either get a bit of scaremongering. We're looking, it's going to be really hard if you buy anyone outside of our ecosystem. It's going to be not a great experience. The walled garden approach. Yeah, it's not going to work.

That's not true. I mean, if anything, now is a great time with APIs and NTP servers. Buy the best of breed that solves your problem and they're all going to be able to talk to each other. So don't fall for that. That's the only risk I see. is the consolidation now means there's a bit more, we're back to that place again. I think we went through this cycle where there's once upon a time centralized ERP and HCMs going, buy everything from us. And that changed into buying best of breed.

I just don't want it to go back there again because we don't need it. So that's my only worry with what's happening. It's probably a habit medium because I think a lot of the organizations that we work with have found themselves, they'll have six or seven different learning platforms. And it's kind of got slightly out of hand. And I think some of this is driven by procurement. It's kind of this kind of procurement piece. We want to have... fewer suppliers,

shorter preferred supplier lists so we can get everything from one place, so much the better. But you're right, I mean, with things like integrations, APIs, you can still go and find that kind of smaller niche supplier, the kind of edgy one. You're going to gain early. It's probably cheaper. And then tie it into an existing ecosystem rather than necessarily having to become as part of a bundle. So now as we look ahead to next year, 2026.

In case anyone's not familiar with what the next year is, we add one to the current year. That's how that works. How do we see AI evolving beyond content production towards real skill development? And Nelson, you've already given an example of the kind of the... practicing for a one-to-one and getting feedback. What else do you see? Yeah, I think the constant production is a bit of that kind of low-hanging fruit. Let's just replicate the workflow we've got now. But I think...

Highway singing is learning and work. I mean, we've been talking about learning in the flow of work for many, many years now, but I think we're actually probably moving to a place where... Like learning is work, right? Rather than looking at work as an execution problem, work will be looked at as a learning problem. Like I need to do a task. I need to learn how to do that task.

I learn and then I do that task. It's all one and the same. And so I think learning will very much start with work. And by that, I mean your work, concepts and data. like I mentioned, the calendar data, but you could have sales CRN data, your help desk data. It could be your coding. It could be what you're deploying to, pushing out to GitHub. All of your work data is going to help me.

figure out what your priority area is that you need to work on to be able to do the work you need to do. Combined with your talent data, that tells me where you want to get to from a career perspective. So that work data will determine what you need to learn. And that learning experience itself... I think it will be as much these kind of...

predetermined trained programming courses as it will be something that's on the fly, self-generated by AI using the context of all of this content, right? So using... MindTools' incredible expertise, but then repurposed into... Then repurposed into a... practice scenario that's super tailored for this particular individual and their work concept it will go through that and then the output will also be it will end in better work right so you're constantly monitoring uh business data and work data

And using that work data will infer using AI whether this person's skills have got better, have you built the capabilities you need, are they improving in performance? So I think AI will help us seamlessly.

Start with work and end in better work and take care of the messy bits in between. And L&D's role will very much be around... designing and architecting that ecosystem where this sits in and i think it's incredibly exciting for that one like i love the idea of we talk about it internally around like if you could give Throughout time, there's great examples of having...

you know incredible teachers who unlocked potential in in somewhere right like i use the example of kind of bruce lee and it man and you know what what it did to bruce lee to where like and and for bruce lee to become who he was and not all of us get that

great teacher who taps that into us and i think right now we have a potential we're not there yet but there's an exciting opportunity to give every single person at work their own it man to not turn you into, you know, a incredible martial artist, but.

to kind of really tap into what is possible with you. And I think that's what it looks like for us. And it's probably not that far away. I mean, we're talking a matter of a couple of years to really refine and deliver that. Yeah, I think there's like the AI that you're describing or those kind of like... It's an augmentation piece, like augmenting current capability by helping you do something that you couldn't currently do.

Some of that is through you learn something new. And some of it is just that the AI does it for you. It means that you can do something that previously you couldn't. But I think you've still got that kind of underlying, and this is where I think the formal instruction piece comes on, something like a workshop. uh, or, um, I do think I like personalized course still, um, gives you that, like, uh,

vast jump in capability in quite a short period of time. So you kind of want this sort of ongoing skill improvement with these periodic injections of like sort of this turbocharged experience. 100%. A great example of it even internally for us is our engineers. to get them onto um ai coding for to get them from the zero to one to understand the frameworks and foundations of how to do it a

a structured course or path was the best way of doing it because they were all pretty much learning something that was completely new. And we needed to make sure certain foundations were done in a structured way to get there. And now we work with a lot of tech. companies as customers and they've kind of said the same right like it's for a lot of that zero to one structured work they're still leaning um on on courses and i think so it's not binary right it's not like

A lot of people saying my end of courses is all going to be here. I don't think that's going to be the case. Yeah. In fact, next week's episode of this podcast, we've got our head of product, Dan Potter, and then Gemma from our learning design team talking about, in Dan's case, woodworking. and try to learn woodworking off YouTube, but actually needing to speak to a master woodworker.

and spend a week with him. And Gemma talking about swimming instruction versus watching YouTube videos on swimming technique. And there's some nice examples in next week's episode that listeners can check out. Ross, you can say something there. Yeah, I think the courses, I think we think about e-learning specifically for a moment. I think a lot of that innovation...

And regarding AI in the e-learning space has been in the back end. It's been how do you make it easier for learning designers to scale content faster? The front end experience hasn't changed all that much. I think the table stakes. for AI and L&Ds for this sort of 50% comes in where people are using it as part of their workflows. That is that, how do you use it to do what we've already been doing, but faster and more efficiently, which is good. But I think...

I think what we'll start to see more of, there are already examples of this out there, is the kinds of experiences you guys have been describing. So where you're taking something from that's highly contextualized and allowing people to practice in the flow of work.

So that's what I'm looking out for next year. Yeah, I think the other thing to look out for is the credibility in the expertise that underpins. the experience where even if it's delivered by AI really matters. Are you implying that this field has been full of charlatans who don't know what they're talking about? Would I say that? No. I think the problem right now...

There's so much focus on the delivery. We think we could be making the mistake of thinking just going to a general purpose AI assistant and getting an answer that we don't know where it's coming from. The reality is if you go to it... track gpt and gemini right now and you ask you know how best to tackle this one-to-one it's going to give you an answer but you don't know what that answer is grounded in why it's grounded in in in the data they've trained this on and so it's not referenceable

It might not be credible sources. You might be leading on, you know, there's going to be a point in time where it is the AI slot we talk about, which is essentially driving the outcomes that you're going to come out with versus knowing. And again, going back to mind tools here, knowing you've got a credible partner who's aligned to your organization and you know exactly where the source of that is.

I think makes a massive difference moving forward. And so it comes in, to me, it's two halves. One is the experience created by AI. Great. but it has to be underpinned by credible referenceable IP and expertise.

I know this is going to be a self-defeating comment. I'm not sure I totally agree with you, Nelson, but I think it's a longer conversation than that because I think that the chat you could hear, Gemma, have become so good that the kind of the hallucination concerns have almost, you don't hear them.

referenced quite so much anymore. So, you know, MyMintos does have an AI assistant built into the platform and you can ask questions and interact with their content library. But that's not really like a special feature now. That's like, I think that is. Ross used to phrase table stakes. I think it's expected. And I actually think, you know, Nelson, you were talking earlier on about you were surprised that the L&D leader is actively using AI.

was only now just over 50%. I think you're absolutely right when you said they're probably not thinking about it in terms of every product that they're using has some sort of AI interface at this point. I think to your point, the hallucinations is less of a... I mean, there's a lot of debate in the tech field around, you know, is it a feature? Is it not? You kind of...

you'll never really eliminate the hallucination completely other than obviously by when you're referencing and undefended by credibility. But there's a huge difference between, let's go back to my Bruce Lee example, right? learning let's say if i could uh martial arts from bruce lee versus um an amalgamation from me yeah that's right from you from you right and and right now What you're getting is you might be using expertise from credible sources, but the reality is you don't know.

like you don't know and the more you start to get into the territory of how a model is going to monetize and how that experience is going to be monetized now we're getting into territory like there was a little period of time i mean even Now, if you go on to ChatQVT, for example, and you ask a very generic question, even ask what are the best restaurants in London, and you...

go through and you look at the kind of top four links that you would get in a search result page from a TripAdvisor and let's say Google WebEase, that's kind of what you're getting in your results. I think there is an element of knowing whether something is aligned to your way of working as an organization. And I think that's why there's an audience for something like perplexity.

I bet it's really perplexity. One of the fundamental values is the fact that something is reliable and credible. Yeah, I know the incentivization of the AI platforms is coming. Rapidly, I'm sure. I'll link to what that term means if people aren't familiar with it. I think we're going to wrap up there because we've hit time.

Thank you, Nelson, for sharing your insights into the past year, but also looking forward to next year. For now, we're going to move on to a regular feature of what I learned this week, where we share something we've picked up over the past seven days. Ross D. What have you learned? Well, we've been talking about AI slop. This week I discovered the Instagram account Learning with Lyrics. Have either of you seen this? Not familiar with that.

Let me try and share my screen. I don't know if this will work. I'll play one of the videos. So this one is about the formation of contrails, which is... white lines you see off the back of airplanes in the sky. That is genius. That is, yeah. There's a conspiracy theory about controls, isn't there? That we don't need to get into. Yeah, I think there's a... People can, in fact, don't Google it. That's my recommendation. Don't Google it. Anyways, I discovered that through the hard fork.

which is a New York Times podcast on technology. I don't really know how I feel about it. I kind of like it. It is AI slop in some ways, but it's also, I don't know. I kind of do feel like I've learned something, weirdly. Yeah, for sure. Catchy, memorable. Learned how... It's a kind of Taylor Swiftification of learning content. She Swiftified everything else.

Nelson, what have you learned this week? This is a shout out to InfoSec lead, Emmerich, who told me about this thing called managed subscriptions that sits in the sidebar of your Google inbox. which essentially helps you unsubscribe from all of your marketing email cons and spam that you don't want in one place and essentially clean up your inbox and make it so much lighter. And it was a bit of a rip.

revelation when i got that one well yeah well that's probably the most useful thing anyone's going to hear on this podcast that's a top tip i'm going to do that now so speaking of ai slop um Long-term listeners will know that I published a science fiction novel last year, Centauri's Shadow, Ross Groening. The book is not written by AI. I wrote it myself. It's all the worst for it. You all know I have a very strict policy about not promoting my book on this podcast.

because I don't think it's appropriate to discuss how many five-star reviews it's received on Amazon, especially in the run-up to Christmas when people are looking at Christmas gifts. It's important to keep things delineated. I did decide to use some funky new AI tools to create a teaser trailer for my book. So I used the Meta AI app to create video clips of certain scenes and I used 11 labs to do the voiceover.

I found a website that takes your book cover and pastes it onto a 3D render of a book. So it's a 3D... I think at the end, and then I used the Splice video app with the Shutterstock Music Library feature to add a soundtrack to it. And the clip lasts about 30 seconds. I'll link to it in the show notes for anyone who's interested. I know Ross D has seen it because he sent me criticism. For my sins.

How long do you think it took me to create that 30-second clip that you've seen, Rostecke? Well, you were off sick all last week, so I assume a week. In working hours, though. In working hours, I'm going to say one hour. It took me six hours. over the course of a week of noodling back and forth. And it was a really nice learning experience for me because we mentioned earlier on this sort of AI equals magic kind of vibe. It is not magic.

really hard to get a decent output from these just particularly if you're wanting something that's super specific because the output is so unpredictable so it took a staggering number of attempts to get something usable so for example i wanted a five second clip of a spaceship at the start of my clip, advertising Centauri's shadow. And I had to generate about 20 videos to get one that looked right. I'm sure it's going to get faster in time. But it's just a reminder that AI is amazing.

It's not just a five-second job if you want to create a teaser for your books in Torrey's Shadow that people can buy if they want in the run-up to Christmas as a gift for all of their relatives. Remember, it was generated by Roth, though. Before clicking that participant. Would that it had been done by AI, it would be better. Despite what all those reviewers see on Amazon.

And that's it. That is all from us. If you'd like to get in touch with us here at MindTools Kineo, you can contact us at custom at mindtools.com or visit our website at mindtools.com. That website will be changing in time, but mindtools.com is for the others now. Our team focused on learning, skill development, behavior change, and performance management, all with the ultimate goal of delivering measurable business impact. So if those sound like priorities for you in 2026,

And if you're listening to this podcast, I can't imagine that they don't, then please do get in touch. And of course, you can work with our partner, how now? Nelson, where can people find out more about? You and the works at HowNow. You can find out more on gethownow.com. And if you do want to get in touch more, I'm happy for you to shoot me a message. I'm Nelson at gethownow.com. And you can find me on LinkedIn too. Excellent.

One final plug. I know many of you, well, some of you, have Mind Tools as your top podcast of the year on your Spotify Unwrapped. And now I know this because I've been sent screenshots of the image. I was able to see it. So thank you very much. If you are a MindTools L&D podcast super fan, you know who you are.

And if you'd like to say thank you to the team who produced the show, then please do share the podcast with others or leave us a review. Both will help us reach a new audience in 2026. We only have two episodes left coming out this year after this one, including our much-anticipated Christmas special. Very exciting. Until then, thank you for listening this week and bye for now.