Key takeaways:

Created by CleanVoice

• Exploratory learning driven by curiosity often outperforms formal training – the traditional months-long course creation model is less effective than allowing people to learn through guided discovery and AI-powered exploration
  
  
• Legacy content uploaded to AI systems won’t drive transformation – organisations need decentralised, continuously curated information models rather than simply digitising old courses
  
  
• L&D professionals must shift from content creation to strategic business partners – success requires focusing on performance consulting and leveraging uniquely human skills like creativity and emotional intelligence that AI cannot replicate

Podcast summary:

Created by CleanVoice and edited by ChatGPT

Jo Cook, Erica Farmer, and Josh Bersin explore how AI is reshaping learning and development from slow, static courses to fast, dynamic support. Bersin argues the ChatGPT wave proves people want answers, not just programmes: exploratory, on-demand learning can outperform traditional training. AI becomes a thought partner, coach, and role-play tool, generating tailored content by role, level, format, and language.

They spotlight Galileo Learn, which turned years of HR content into hundreds of interactive, tutoring-style experiences. But they’re clear: dumping old PDFs into AI doesn’t fix outdated thinking. Content needs constant curation and smarter, decentralised ownership, with regions updating material while central teams maintain standards. The message for L&D: stop defining success by courses, completions, and roadmaps. The real job is solving business problems, driving performance, and staying responsive in a world where fixed plans age fast.

Links from the podcast:

TJ’s 60th anniversary conference

Thanks to ⁠TechSmith⁠ for providing Camtasia and Audiate for editing

Speakers:

1. Erica Farmer
2. Josh Bersin
3. Laura Overton

Transcript:

Created by CleanVoice

Jo Cook:
Welcome to the Training Journal podcast. I’m Editor of Training Journal and your host, Jo Cook. How we become business leaders,

Laura Overton:
We become business leaders, how we have influence in our organisations to help others become equipped and ready in a world of work that we just haven’t experienced before.

Josh Bersin:
Probably 70 -80 % of the learning situations, people don’t want a course.

Erica Farmer:
The great thing about artificial intelligence is we can use it in so many different ways.

Jo Cook:
In this episode, we focus on AI’s use within L &D, both to make our offerings better than ever before, and from a strategic perspective to improve organisational outcomes. I first start discussing the chat GPT boom with L &D HR and business analyst Josh Burson, and what L &D can take away from the change in our industry.

Josh Bersin:
Two things I think we can learn from that. Number one is that human beings are by nature curious animals. And I think this gets back to this idea of curiosity. I mean, there’s sometimes when you have to learn something and you don’t know where to start and you want a scaffolding and a structured course and you want to sit down for a while and you really want to be taught because you just don’t even know enough to absorb the material. But after a while, when you get comfortable with a domain, you don’t need that as much, maybe once every year or two. But you want to dig around. You want to try things. You want to experiment. You want to learn things that you know you needed to do better. Or maybe day -to -day, you uncover problems. And I think ChatGPT proves that that use case of exploratory learning might be 100 times bigger than the formal learning we thought we were doing all these years. Now, I can’t exactly predict that number, but that’s number one.

And the way I frame this is that most of the technology -based tools and solutions we’ve built in learning over the years have been replicating what I call the publishing model. The publishing model is I’m an expert or an instructional designer or a training manager or a training department. I’m going to study the problem and the domain. I’m going to build a course or a program. do all sorts of great stuff. I’m going to put videos in there and interactions and simulations. I’m going to make it really fun and interesting and I’m going to give it chapters. And then once I get done building this thing, it’s going to take me a while to build it. It might take me six months. Then I’m going to launch it and you guys are going to love it. And if you don’t love it, I’m going to be really unhappy. So I’m going to measure how much you love it and I’m going to do a Kirkpatrick or whatever. And I’m going to do an ROI analysis and then give you guys all copies of it. And I’ll let other people translate it for me. And then eventually I’ll find different versions of this around the company. Well, that paradigm is very problematic. It takes a long time. It’s complex and slow and expensive to build it. You can’t translate it into multiple languages or multiple formats very easily. The individuals who already know. The first 10 chapters are just going to be frustrated with the whole thing, or maybe they’ll find their way to the part they want, but maybe they won’t. And it’s a big legacy object because once you finish building it, this thing, this course is kind of a big blob that you have to maintain. And then you get a whole bunch of those and you don’t really know what their utility is because you don’t really know who’s using them because the LMS doesn’t really know what’s in this. course.

Jo Cook:
Great point, Josh, about how new technology is often used to replicate previous models of work, albeit more efficiently. Now, this is fine in some cases. A tractor is an easier and quicker way to plough a field than using a horse. But a tractor actually gives you more. But a tractor actually gives you more. Like it can power other machines on the farm too. It can use hydraulics to lift heavy items. It can be upgraded or modified with new technology. All of a sudden, the technology isn’t just doing the same job more efficiently, it’s offering new ways of working. We’ve all heard those phrases like AI is the worst it’s ever going to be, and we know that development is so super fast and things are changing quite frankly quicker than most of us can keep up. Erica Farmer, co -founder of Quantum Rise Talent Group, has this to say about AI’s flexibility.

Erica Farmer:
The great thing about artificial intelligence is… We can use it in so many different ways. So evaluation is a big part of it, depending on what problem you’re looking to solve and what you’re actually evaluating. But you can use it as a thought partner, give it the prompt that it’s an evaluation specialist and this is the kind of data that you’re going to feed it, ask the narrative that you want from it, get it to create models that you need, what’s the story that you need to tell your stakeholders, any challenges, any blind spots, all that kind of great stuff. Really using large language models as another thought partner, kind of ideation, critical thinking, all this good stuff is perfect. So using it as another trainer, facilitator. ability for delegates to role play with it, practice new skills. If you’re having that difficult conversation, if you need to practice influencing or pitching or presenting or whatever it might be, you can do that with an LLM. And I’m not just talking about typing into it. You can speak into it on your phone, on a mic, whatever it might be. And we can use all of this as opportunities for a thinking is another digital colleague, a trainer, a coach, an actor to really embed those skills going forward.

Jo Cook:
Josh continued this theme about using AI and L &D and looking at this from the perspective of the publishing model.

Josh Bersin:
Well, if you look at what ChatGPT does, look at what ChatGPT does, if you generated that content dynamically, in other words, you took the subject matter expert information either from a book or a recording or an interview or whatever document you have, the system can dynamically build content that will… meet the learner’s needs based on their situation, based on how advanced they are, how junior they are, what phone or interaction, or do they want to talk to the course, do they want to read, do they want to take a video, whatever. So this idea of a dynamically generated content system that can be filled with content faster and faster and faster and can generate instructional content, I don’t mean any content, but instructional content in multiple languages, in multiple formats. is a massive paradigm change. It isn’t using AI to do the old stuff faster. It’s doing it differently. And the reason I’m so high on this is because I think this human dimension of curiosity, to me at least, is a proof point that in many situations, probably 70, 80 % of the learning situations, people don’t want a course. They want the answer to a question. They want a podcast. They want to spend 10 minutes with an expert or they want to be quizzed or they want to watch a video, whatever it may be that’s most appealing to them. And in all the years I’ve been doing this, I’ve run into so many use cases. I mean, the one that I remember that’s most vivid that I always kind of laugh at, it was a long time ago, but British Telecom was teaching their repair technicians how to repair all the various phone systems in the UK. There’s so much legacy stuff. I mean, this is the way the phone industry is. And so what would happen is somebody would be highly trained. They’d go out into either a business or an individual’s house and they’d open the wall and they look at the phone system and they’re like, I don’t even know what this thing is. I’ve never seen it before. So they say, well, we can’t train everybody on every piece of phone equipment we’ve ever sold because it’s just impossible. Plus, we don’t even know what this is, really. So why don’t we have them snap a photo of the thing they found and email it? to corporate, and then we’ll tell them what we think it is and how to fix it. So that was the beginning of this. But imagine a situation now where this person takes and snaps a photo, sends it to the dynamic content system, dynamic content system recognizes it, sends them the parts manual, sends them the video, and they interact with the content just like ChatGPT to figure out what to do. How do I start this thing? Where’s the power supply? Whatever. I mean, There’s thousands of use cases like this, and you can apply that to soft skills. You can apply that to technical training, sales process, all sorts of things. And the reason that I’m high on this is because we’ve done it. We have about eight years of formally designed traditional content in HR about leadership and management and recruiting and all that. It was really good stuff, but it was getting a little dated. We weren’t able to keep it up to date. We moved it all over into our AI platform, which is called Galileo Learn, but it’s an AI dynamic content system in five months. We got the whole thing in there. We now have 700 smaller courses and you can go into each one of these courses. I’m not sure whether we should call them courses, frankly, but, and you can ask it questions. The system will tutor you. At the end of the course, there’s an AI Josh that comes in and gives you a big quiz and sort of interrogates you. I mean, this is so exciting to me. And there’s several implications of this. Number one, of course, it’s faster and more effective in building content. So that’s a cost part of the L &D function. But the bigger benefit to me is that the consumer experience of the employee or the manager or the worker trying to learn something. They can find what they need. They can get their questions answered more quickly. They can advance as fast or as much as they want.

Jo Cook:
And this is broader than just learning professionals, professionals, right?

Josh Bersin:
The subject matter experts who are not in L &D, who are in the business, can use the system themselves to load content or update it so that it’s current. And we don’t have this problem of L &D constantly doing performance consulting projects. and iterating. So I think there’s going to become a much more decentralized, federated approach to learning and development where a lot of it will be done regionally by sales teams or marketing teams or manufacturing teams. And then the corporate group can create standards and tools and other artifacts that support this. And that’s been a problem from the beginning because early in my career, the most sort of iconic research I did in around 2004 or so, was what I called the High Impact Learning Organization. Because in the early days of e -learning, nobody could figure out how to manage all the e -learning stuff that people were building. What I learned, at least my experience, was that you can’t centralize this function. It’s too local. So you need a process and sort of an operating model where you can build infrastructure and tools and content globally, but also regional people. can use the learning infrastructure for the thing that just happened yesterday that’s maybe new, that nobody knew about, and we don’t have time to go build a course on it. And I think this is going to facilitate that.

Jo Cook:
One thing you mentioned in this information was you were saying about, you know, legacy courses, they become out of date, they become hard to manage, et cetera, which I completely agree with. But also… that we’re using some of that and we’re putting that into our AI tool of choice. So at what point is that still the same problem that the PDF we uploaded, the video of fixing the PABX system, whatever it is, that video can still become out of date. That PDF can still be wrong. So is that not the same problem?

Josh Bersin:
We’re going to have that problem again. I mean, we’re experiencing this in our company. So we have 25 years of research in our AI. and some of it’s old. And so frankly, we haven’t been doing enough of this, but we probably need to go in there and take stuff out that’s out of date, vendors that are gone or things that are not relevant anymore. There’s still going to be content management, but it’s more of a general purpose content management problem than course by course, update, edit. it’s going to be much easier. And by the way, and I think it will have many more tools because now we’re talking about generic knowledge management, which every company has in all sorts of domains. So you could go to the AI and say, show me all of the references you have to part number 2235. You could go through and say, okay, please delete all references to part 2235. We don’t sell that anymore. and tell it what to do.

Jo Cook:
And so this is obviously an area of huge opportunity and for some people, huge concern, like you say. Is there still going to be this meaningful role for L &D? Because some people like using Camtasia to edit their videos. Some people like making their PowerPoint decks. Not everybody wants to analyze the data or focus on the problem. So how can we meaningfully still work in the age of this super worker that you’re describing? Well,

Josh Bersin:
I used to love my Selectric typewriter too. I mean, I really got a kick out of it when I was at IBM in the 1980s, but it’s gone and unfortunately I can’t use it anymore. So, you know, unfortunately or fortunately, you’re going to have to upgrade your tool set. Things that you learned how to do in the tools that you use are still going to be useful. in the new tool set. And you’re probably going to find there were things you did in the old tools that you can’t do very well in the new tools. So you’ll be using the new tools for 80 % of the work or 90 % of the work, and then you’re going to have to add stuff to it. And that’s just the way AI is, because AI is so new. But there are things you can do in the AI that you couldn’t do before. You know, like, for example, I mean, just to use the Camtasia example, if you want to build captions on a video, You can do it manually in Camtasia, or you can upload it to an AI and don’t do anything. And just let it do the thing, let it do it, and let it translate it into 22 languages for you. One of the questions you asked I think still is important. Is there a role for an instructional design function? I think there is. I think the AI is pretty good at creating what it thinks is instruction. But humans know more. I mean, humans are infinitely more scalable and infinitely more intelligent than AI. That’s my feeling about it. So you know what works in your company. You know what’s going on in this business area. I think we’re going to spend a lot more time with the business doing what we might call performance consulting and needs analysis and a lot less time, you know, authoring. And if you’re really creative and you love to author great stuff. You can still do that. You’re just going to have better tools and more interesting tools to do that with. You can create beautiful avatars or beautiful scenes or really powerful simulations. I think everybody in L &D is going to have lots and lots of things to do. It’s just going to be different.

Jo Cook:
You have to move with the times. You have to pivot. You have to change. Otherwise, you become obsolete yourself. I was speaking to Laura Overton this morning, and we were talking about the enablers which she’s focusing on in her book. And she says, is there a danger of L &D professionals picking up a new name, a new mantra, but actually not changing our approach at all? A bit like you were saying earlier on about the same thing.

Josh Bersin:
There is. And here’s the paradigm problem. And I saw this when I first got into the space. Some people in L &D are teachers and they want to teach and they like teaching. And they don’t want the AI to do the teaching because they don’t trust it. And I’m sure most people in universities and schools are feeling this way right now. And what I think most teachers are recognizing is that you can still teach and the AI can be your assistant. We call it the super worker effect. The super worker effect is everything you know how to do, you’re still way ahead of the AI. But let the AI do more and more and more of the stuff. that you might consider to be routine. And so if you are a teacher and you like standing up in front of the classroom and lecturing and entertaining people and all that, fine, there’s still a role for that, but you’re going to have to fit that into the enablement of these AI tools. I think there will be some jobs that go away. Like Bill, my business partner used to be the CLO at Deloitte. And he told me when he was a CLO at Deloitte, He had something like 100 people in India that just loaded courses into the LMS around Deloitte, around the world. I don’t think those 100 people are going to have jobs. I don’t think that kind of routine work is going to exist. They’ll have to find other things to do, but that kind of stuff nobody really likes doing. In some sense, you can now think about your teaching skills as teaching the AI how to teach other people better.

Jo Cook:
I mentioned Laura Overton in a conversation with Josh. Laura has a huge background in learning, innovation and business impact. So she’s ideally placed a comment on this point about changing roles in L&D.

Laura Overton:
When the business starts to question whether or not we are valuable or doesn’t even consider if we’re valuable, then we’re in a really dangerous zone. But when the business wants us to be able to deliver business value and we see that we have an opportunity to do that. That’s when we can start working together, co -creating value together. When we’re working at the business, there are so many things that actually work together in our situation, work against us. There’s constant change in the business, constant shifting of priorities, constant interwoven actions of individuals that get in our way of delivering against our traditional roadmap. We want to explore how to operate in uncertainty, how to currently show up to business leaders and why we need to shift. We want to explore what it means to be a commercial enabler, how to shift our language, to speak business language. And by driving business value, we raise ourselves up as the business leaders we need to be.

Jo Cook:
It’s interesting, the comment about operating in uncertainty. As part of the planning for Training Journal’s 60th anniversary conference, I was looking back in the magazine archives. This is a quote from Roger Daw, Chief Executive of Manpower Services Commission, in 1984. We are in a period of unprecedented change in our industrial and occupational structure and in the nature of many jobs. This presents a great challenge to all of us involved with training if we are to meet the needs of employers and individuals for the skills required for tomorrow’s jobs. End quote. That could have been written today and quite frankly, at any point before or since 1984. With this in mind, let’s head back to the discussion with Josh. How do you see human intelligence and all of those people skills, soft skills, whatever people would like to call those, working alongside AI, especially when, you know, if we look into a year or three or five years into the future, when AI is so much more fully integrated into organisations?

Josh Bersin:
Well, you know, there’s a lot of talk about artificial superintelligence right now, but I think it’s a little bit misleading, and I think the tech vendors are way, way over the top. Because human intelligence is a genetic combination of billions and billions of genes that have been built and developed and evolved over millions of years of evolution. So every single one of us… has the learning experience of millions of people before us, including the epigenetics of our experience in our lives and our parents’ lives. We adapt very fast. We create, we innovate, we are curious, we have emotions. I call it the unquenchable power of the human spirit. AI doesn’t have any of this. AI doesn’t have ingenuity. It’s really good at doing stuff that you know, it’s good at. But I don’t think it’s as human like as the engineering, you know, AI guys say it is. And I see this all the time. I mean, you know, what makes a business successful? It isn’t routine, highly productive agents. It’s new ideas. It’s customer relationships. It’s innovation. It’s understanding the market. It’s seeing around corners. It’s taking risks. There are all sorts of human things. It’s empathy, it’s sympathy, it’s love, it’s caring. These are human things. I think we can mimic them in AI, but I don’t think AI has the genetic characteristics of humans. I’m not a mathematician, but I just don’t think they’re ever going to get there from here.

Jo Cook:
And it can help us make those decisions. So when we’re talking about horizon scanning or looking around the corner, it can help. Number crunch, you can look at the data. Yeah, I mean, you can take a spreadsheet and put it in an AI and just ask it questions instead of doing pivot tables and spending three hours trying to figure out how to analyze the data.

Josh Bersin:
I mean, spreadsheet and put and put it in an AI and just ask it questions instead of doing pivot tables and spending three hours trying to figure out how to analyze the data. I mean, and then that makes you as a human more insightful about what the situation may be in your business or whatever. That’s why I call it the super worker effect. If you use AI well and you’re not intimidated by it, you’re going to be smarter and better and more high performing and having more fun in your job. And you’re going to be delegating some of this junk work you didn’t like doing anyway over there to the AI guy. And so I have a very positive feeling about it.

Jo Cook:
One last question for you is around the curiosity element, which I think you’re absolutely right. It’s so important. And what about if… We have these cultures where we ask the questions of the AI. It’s doing the processing and the analytical thinking. Are we going to lose this agency we have around curiosity? How do we have our learning cultures in life and also in our organizations to counter that?

Josh Bersin:
No, I absolutely don’t believe we will. And here’s why. I have two grandchildren. One of them is two. The other one’s about three months old. And I have had a chance to, you know, learn again what children are like. They are constantly curious. They are constantly learning. Everything is a learning experience to them all day, every day. Read and write and understand what things are. And, you know, if you want to teach your children about something, you just tell them a story and all of a sudden they’re fascinated. That is a genetic characteristic of human beings. sort of the king of the beasts. I don’t know exactly what to call it, but there is this sort of nature of learning and evolution. That’s why I call it the unquenchable power of the human spirit. It’s almost religious to think about it. Now, I think there are situations where people stop learning. They get frustrated. They’re imprisoned in their own minds. They have mental illness. They’ve been abused. But I think most of us can learn our way out of those problems if we’re given the right support. And a lot of our jobs in HR is supporting people to be the best versions of themselves in a psychologically safe environment. And when that’s done well, and that’s what my whole book, Irresistible, was about, when that’s done well, people outperform your expectations all the time. And I think that’s just the nature of human beings. Sometimes they outperform in strange ways that are a little inconsistent with others, but that’s why we have coaching. That’s why we have companies. But I don’t think we’re going to lose our creative instincts whatsoever. I think we’re going to free up more creative thinking because we can delegate some of this distracting stuff to these agents.

Jo Cook:
Of course, this means we’re looking at learning and performance development as much bigger than instructional design. So who better to have the last thought than Laura Overton?

Laura Overton:
We’re working on what matters to the business rather than what business might expect from its courses and its program. And I’ve seen that those operating for the last 20 years at the business value end of the spectrum are increasingly absorbed into business, become increasingly valuable in business. And it’s where we want to be, solving business problems. The challenge is that most of us have been measuring and reporting on learning metrics in the past. Something that’s so important to me is how we become business leaders, how we have influence in our organisations to help others become equipped and ready in a world of work that we just haven’t experienced before. I think one of the things… that I’ve seen in my own research is for decades, literally, we have been operated with roadmap, with models and our tools and our technologies, detailed plans and tracked and predetermined destinations to help us get better business impact. But the problem is, is that that map is out of date before it’s even printed. And that’s why it’s so essential that we become. business leaders in our organisation rather than the learning leaders that we’ve traditionally been used to being. Traditionally many individuals have seen their value very much at the learning value end of this spectrum but on the other end of the spectrum what we’re able to see is organisations and individuals who are delivering business value who see themselves as contributing to business value, to performance and to competency.

---

Follow the Training Journal podcast on your app of choice, or sign up to the newsletter to always be in the know!