Artificial Intelligence has become a major part of our daily lives. From applications in search engines to online shopping and streaming services, AI is making our lives more efficient.
While AI continues to evolve in its sophistication, many L&D functions are yet to incorporate it into their learning programs. Fortunately, my guest this week is here to offer guidance on how you can apply AI models to your L&D practice and what the best use cases for AI in content creation and data analysis are.
In this podcast recap (check out the full episode here), I speak with Donald Clark, CEO at WildFire Learning, as we explore five applications of AI models in L&D. As an award-winning speaker, author of the books ‘AI for Learning’ and ‘Learning Experience Design’ and co-host on the podcast ‘Great Minds on Learning,’ David is the go-to person for practical advice on how to use AI to increase efficiency and help employees achieve their learning goals more effectively.
So, what is AI and why should you be thinking about its applications in your L&D practice? Read on to hear more.
As Donald explains, AI is important because it’s the technology of the age.
“This is a form of technology that is not yet there,” he says, “but it looks as though it's heading in a direction which transcends the abilities of our brains and cognition, and it already has in several areas.”
AI has huge implications not only for us as a species, but for every area of human endeavor; research, science, solving climate change, and not least for work and employment.
AI mediates all your social media, everything you buy on Amazon, Uber, and what you watch on Netflix. “And yet many L&D teams struggle to incorporate AI because it doesn’t typically fit into existing frameworks used to build training courses,” says Donald.
But before we jump into some of the use cases of AI in an L&D context, let’s start with a definition of what AI is.
Looking for more tips on how to adapt to new ways of working in a digital world? Download our cheat sheet: 5 Secrets to Keep Your Teams Engaged in the World of Hybrid Work
As Donald explains, there's no formal mathematical definition of AI, but essentially, it's just smart software.
“First of all,” he says, “we can’t say this is just math and software because AI is competence, but without comprehension. There is no self-consciousness or brain behind it—we must remember that AI does have the same intelligence as the human brain.”
In Donald’s experience, there is one simple way of summing up AI for the L&D function— it’s very good at precision tasks. “AI only knows its task and as long as we understand that, we can start to make progress.”
Sounds exciting, right? But why should you be thinking about incorporating AI into your L&D practice? Read on to hear why.
When it comes to AI in L&D, Donald says it’s already having great success with adaptive learning.
Donald goes on to give us an example of how adaptive learning is making an impact with a system he helped to build that has just been bought by Cambridge University.
The system broke content down into individual components across the curriculum. Once we broke the content down, we could allow users to vector through that content in different ways.
“So if you have prerequisite knowledge, the system would say: hold on, you need to cover this and this before you move onto that, because, en masse, it just doesn't work,” Donald says. “So, I think adaptive learning is one area where there has already been success and will continue to see success in the future.”
Donald finds that another big area for AI is doing away with the three-day classroom course which is then imported online.
“It's only now that we're beginning to see technology that will allow us to do performance support in a very smart way,” he explains. “And this is what I call smart content.”
“People learn more informally now—they drive their own learning needs and learn as they go about their daily lives. There are some L&D leaders who are adapting their learning programs to suit the learners of today.
But many L&D teams still struggle to shift from the traditional mindset of delivering courses in what Charles Jennings calls the ‘conspiracy of convenience’ where managers simply request courses because that’s what L&D departments have always done.”
The scene is set, but where to from here? Read on to hear about five AI models with learning and development potential.
There is plenty of AI software that already exists, but as Donald explains, the following have applications within the L&D context.
A new area that few people are aware of, these AI models called transformers are great at breaking content into a number of little components.
“First of all, these transformers are pretty good at summarizing. For example, they will take Romeo and Juliet and summarize it into a paragraph. And that paragraph will be really good,” says Donald.
“In an L&D context, it looks a little something like this: every training department receives large PDFs of 10,000–12,000 words, and you have to turn it into a training course. You can use a summarizer, defining how many words you want the description to be whether that be a hundred words, a thousand words, or ten words, and it will give you a summarized version.”
In Donald’s experience, summarizing is a really important aspect of AI which can be really effective for L&D content purposes.
In addition to summarizing texts, you can create texts.
“People are using this in marketing all the time now,” Donald explains. “These systems work by prompt. So you can say, ‘Today we are very excited to announce…’ and the transformer will write a piece of marking copy that will be very crisp and it will carry on predicting the next word you need.”
“You can create questions, and answers to questions which significantly improves efficiency because typically, for L&D teams, this task has been very time consuming,” he says.
You can also create graphics from text.
“So the first very famous example that came from DALL-E originally was an astronaut riding a horse in a photo-realistic style. Now that is an astonishing challenge in AI, which has now been solved–you can create graphics from text.
“Not only can you create the graphics from the text, but you can also say, ‘I want a duck in the corner’ and it'll put a duck in the corner. And all the lighting in the photograph or the image will adjust to that.”
Again, this is a great example of how AI models are transforming the way we build the content for training programs.
Already going for the last decade, speech to text and text to speech is solving the issue of accessibility for many people.
“This model refers to pure AI. And people who have visual or hearing impairments have been using it for a long time.”
Finally, the big one for Donald is also something he’s been heavily involved in for the last three to four years–the creation of content.
“To create content, you simply need to send us a PDF, a video, or a PowerPoint and we use AI to summarize it, telescope it down, and with the touch of a button, it produces much of your learning content.”
“The AI produces open input where you're actually typing text in. You're making the effort cognitively to learn this stuff. And we even have complete open input where AI is used to semantically interpret the answer.”
In Donald’s experience, analysis is important and the decision of whether or not you are going to be doing any training at all is almost a given.
“We've been working on a project for a couple of years now called The Blend. The data analysis method we’ve got is not dissimilar to what we built in the past—a spreadsheet in Excel where you input: how many learners have you got? Are they working at home?” he says.
The process is about gathering good data about the learner, what the learning is, and what people are actually going to be expected to do in terms of performance.
“There's a big difference between riding a bike and learning a piece of theoretical physics, and then you also have things like how effective is this knowledge going to be in terms of pedagogy models–will this information actually transfer or not?” explains Donald.
So, Donald and the team built The Blender which takes all the variables and uses clever AI math that integrates the pedagogic models around transfer. “There are about twelve criteria for good transfer from the literature, and we took that and applied it so that it spits out an optimal blended learning solution,” he says.
“But sometimes that solution is not to do a training course at all because we absolutely know that this more recent learning in the workflow performance support approach is sometimes far more powerful, takes less time to reach competence, saves you a pile of money, and is actually what users want.”
Sometimes that solution is not to do a training course at all because we absolutely know that this more recent learning in the workflow performance support approach is sometimes far more powerful, takes less time to reach competence, saves you a pile of money, and is actually what users want
Once the data is gathered, the decision might be not to do any training because people might not have to remember things. “It's only if you need to really have some sort of cognitive change that you need the more intense approach to training. And even then a course is not necessarily the best solution,” Donald notes.
So, what does the future hold when it comes to AI in an L&D environment?
As Donald explains, we’re already living within the world of AI because the internet itself is AI-driven.
“Almost everything you do online is mediated by AI and quite rightly so because it needs to be automated. We want our lives to be easier so we can do other fruitful things. The impact will be immense because this is the way all technology is going. It's smart content, smart analysis, and smart platforms.”
And, as Donald says, it’s not that what we had was outdated, we just had the stuff that the technology was capable of doing. “There's nothing wrong with the systems we have in place,” Donald says, “but now we have new software that can personalize and do these wonderful things in learning better than ever before.”
Thanks to Donald for sharing his insights and experience on AI-driven learning with us! Keen to learn from more L&D experts? Check out my conversation with Paul Matthews on three analyses for designing impactful learning workflows.
Want more peer insights on transforming workplace learning? Sign up to become a member of the L&D Collective, and check out our other #CLOConnect interviews with top L&D leaders on driving growth and scaling culture through Collaborative Learning. Or you can subscribe (below 👇) to our weekly newsletter to receive our latest posts directly in your inbox.