Start with the human – using AI to supercharge learning design

In IT, Collins felt her work no longer involved neuroscience directly, but when she found herself managing an IT training team, her colleague made a comment that changed the trajectory of Collins’ career.

“I employed a very smart woman to deliver our IT training while I continued to focus on technical support,” recalls Collins. “And she said ‘Stella I can’t believe you don’t want to do this, because it seems right up your street – it’s about psychology and people.’”

That insight nudged Collins onto a path to explore evidence-based learning, to write a book about learning and the brain and ultimately to co-found Stellar Labs, where she is Chief Learning Officer.

The Belgium-based company enables users to build and deliver effective, evidence-based learning pathways on their online platform, and AI is an important tool in their product.

Evidence-based learning
As Collins dug into the psychology and neuroscience of learning in the early and mid-2000s, she homed in on what at the time was called ‘accelerated learning’, which was something of a misnomer, she recalls.

“It was called accelerated learning, but it wasn’t about making learning faster, it was about making learning more effective,” she explains. “Later it became known as brain-friendly learning and today we call it evidence-based learning.”

Whatever we call it, the approach boils down to working with our brains to ensure that people are motivated to learn, and creating a learning environment and design that actually works for people.

Collins likens the process of effective learning in any context to learning how to ride a bicycle.

“You start off learning to ride the bike in a safe place, with someone there to guide you and to help you if you wobble, and you have stabilisers to support you,” she says.

“Then as you build skills and confidence, the stabilisers come off. You still need some guidance, but you are experimenting and going for longer stretches by yourself. Then at some point you move out into the road, and you learn to navigate that real-world environment, and again there is probably someone with you to keep an eye on those first journeys as you build your independence.”

Changing gears
A sequence of motivated learning and scaffolded practice with support is a useful template for workplace training, notes Collins. Stellar Labs has developed a model to provide that path for people as they learn knowledge and skills and start to apply them in the real world.

“You eventually need to move from the workshop, the classroom or the e-learning, to go and apply what you have learned, to do something for real,” says Collins. “It’s a bit scary so you still want some support and feedback and you will still make mistakes but you learn from them too.”

The ultimate success is that based on learning, a person will change their habits and behaviours – whether it’s cycling from A to B or applying new knowledge and skills in their lives and the workplace.

“For this to happen, you need lots of space, support and repetition,” says Collins. “And if you skip bits along the way, you won’t achieve the same success in learning. The vast majority of people who go on a training course don’t get to practise the new skills with support and feedback, so their behaviours probably won’t change in the longer term.”

That’s why Stellar Labs developed the GEAR model for learning design, to support people to make those meaningful changes. The building blocks are Guide, Experiment, Apply and Retain, and the approach is designed to encourage people to adopt, apply and retain the new learning.

“It’s a framework that works with our bodies and brains, and it’s based on how we naturally make connections,” Collins explains. “And the design means people go from knowing to doing, and we see behaviour change.”

Bringing the brain to book
As Collins started to find out more about how our brains learn best, she was inspired to write the book ‘Neuroscience for Learning and Development: How to Apply Neuroscience and Psychology for Improved Learning and Training’, which was first published in 2015 and is now in its third edition.

In it, Collins explores how evidence-based psychology and neuroscience underpin effective and successful training and learning, including in the design and delivery of face-to-face, online and virtual learning.

“What I really liked about writing the book was doing the research to underpin it. That helped me structure my own thinking and to question my assumptions sometimes,” she says.

“And with each edition, I had to go back and revisit the science to see what had changed and the trends, and update the contents accordingly.”

In the course of those editions, Collins has seen an upsurge in technology being applied to learning design, particularly in light of the COVID-19 pandemic.

“Technology was becoming important, but the pandemic really accelerated the accessibility and broad acceptance of technology in learning,” she says. “People were really starting to think much more about how to use technology, because they had to.”

The latest edition of the book coincided with the mass arrival of accessible forms of generative AI, such as ChatGPT, which immediately captured the public imagination and attention. Collins has witnessed AI evolve over the decades, and the technology is now an important tool for Stellar Labs.

“I first came across AI when I was a student in the early 1980s, then I worked with expert systems in the 1990s, which were an early form of AI by another name,” recalls Collins. “In Stellar Labs we have increasingly used AI in a kind of needs-led evolution, and now there has been an explosion in the tools and capabilities that we can use both as a business and to build into our platform.”

AI supercharge
So how can AI supercharge learning design? For Collins, the answer lies in using the technology to make the process of learning design more efficient.

The key is to keep the human at learning at the core, then use AI to do the heavy lifting around implementing the design, she explains.

“What we are about at Stellar Labs is to make sure that every minute spent learning is worth it,” says Collins, who co-founded the company in 2019 with Raf Seymus.

“We use neuroscience to do that, and we are very curious about new science, new technology and now we use AI to make our process more efficient.”

AI means a user can draft a learning journey on Stellar Labs’ platform in as little as 10 minutes, but the focus remains on quality rather than quantity, Collins stresses.

“Our starting point is always on the behaviour we want to change, and then building the learning design around that using what we understand about how the brain learns,” she says.

“The AI is a tool that lets humans build the training more efficiently, it saves time and effort, and you are still creating an effective learning path that can change behaviour, you can see progress and every minute the learner spends is worth it.”

Stellar Labs has in-house technology and AI experts so their platform is nuanced and can deliver the desired results, explains Collins.

“We initially used someone else’s technology, but we found we couldn’t do what we wanted, we couldn’t measure learner progress and behaviour change in the way we needed to,” she says. “Now with our own experts, we can start with the learning and learners first and shape the technology around that.”

Their process starts by focusing on the behaviours that the training will change – and uses AI to source materials and build in crucial features such as scenario tests, spaced repetition and work-based actions.

“The first thing our tech does is ask you what are the behaviours you would expect to see in somebody who has this particular skill, and that’s a great starting point for analysing the learning needs,” she says. “Then the AI can create the content quickly in an effective design that is built with humans in mind.”

Getting personal
Every learner is an individual with their own desired outcomes. Taking a one-size-fits-all approach probably means that the size does not fit most people and situations exactly, and Collins sees AI as an important tool for personalising learning paths for learners, leading to more satisfying outcomes.

“At the moment, we use Large Language Models to generate content, and we are bringing in a feature where organisations can use their own material to feed the AI,” she says. “This means the material is relevant, it has a deeper level of trustworthiness and there is more tailored input for the learner.”

AI can also drill into data from a learner’s journey to deliver insights about how behaviours are changing, or not, she adds.

“We can use AI to help support a learner directly on their journey, based on data analytics, to nudge them positively towards applying knowledge and changing patterns of behaviour.”

However, Collins warns against simply using AI to generate more and more content, which could overload the learner more.

“Learning is effortful, and what we need is for people to be supported, not just to be given more information to learn,” she says.

Trust and cybersecurity, too, are important issues.

“There is evidence that seems to show that people actually sometimes are more comfortable talking to AI than they are to a human trainer,” she says. “So there has to be a trust there, a recognition of the learner’s vulnerability, and their information needs to be protected.”

Live and learn
Research is important to Collins, who is keen to improve how AI is applied to learning, and Stellar Labs is working with researchers at the University of Antwerp on learning analytics to support learners and help organisations meet their goals.

The company is also working locally with IMEC (the Interuniversity Microelectronics Centre) in Belgium, to identify behavioural design techniques optimised by AI.

Collins herself is a frequent expert guest on podcasts, she is an active participant in the Learnovate community and she speaks at conferences on the topic of neuroscience and learning.

In 2022, she delivered a keynote at The Learnovation Summit about harnessing brainwaves for peak performance in a knowledge economy, where she emphasised the connections between brain and body and the need to use both.

And that is why, while AI can undoubtedly make learning design a more efficient process, at the end of the day how we learn is fuelled by the basics, according to Collins, who brings it back to supporting our biology.

“The best things you can do to help yourself learn are to sleep, exercise and eat well for your health,” she says.

“We often think of sleep as a time when our brains are off, but they are not, they are still working away and running processes that don’t happen when we are awake,” she says. “We need these processes to be able to learn effectively.” Exercise too is a brain tonic: “When you are physically active, it supports your brain health and function in many ways, including boosting a molecule called Brain Derived Neurotrophic Factor, or BDNF, which helps us to grow new connections in our brains,” she says.

And increasingly, we are understanding that what (or who) lives in our guts affects our brains too, notes Collins.

“I’m becoming fascinated by the microbiome, and how eating foods high in fibre and gut-friendly nutrients can influence the microbes that live in our gut, which in turn communicate with our brain.”

The bottom line? Look after the engine that drives your own learning.

“We use neuroscience so your learning is fully effective, and AI to improve the efficiency,” says Collins. “Your role in the learning game is to keep your brain in tip top condition.