If We Can Build a Self-Driving Car, How Close Are We to the Self-Building Course? (Part 1/2)
Course creation has never been easier than with “HR’s Best Friend” (yes, it’s an AI). CanopyLAB’s co-founder & CEO Sahra-Josephine Hjorth gives you exclusive insight into the mind-blowing concept of a self-building course.
Rather listen to the entire interview? We’ve got you covered.
If you don’t have a wild idea, you’ll never really achieve something really big.
∼ Sahra-Josephine Hjorth, CanopyLAB’s co-founder & CEO
Question : What does a self-building course look like and how will it benefit users?
“If we can build this self-driving car, imagine all the components that go into that. There is a ton of swarm intelligence ensuring that cars don’t hit other cars. And I said well, if we can build such a complex mixture of hardware and software, then of course we could also build a course that builds itself.”
Before diving into the elements that will make a self-building course possible, Sahra-Josephine takes a step back to review the actual need of automating online course creation. When our sales team engage with clients, we see that 73% of that clients that don’t pick CanopyLAB opt out of digitizing their content or corporate training programs altogether. This is due to the major burden HR managers face when it comes to transforming large amounts of content into engaging digital experiences for employees. Essentially it is too expensive due to the numerous amount of hours it takes, particularly in countries with a high hourly salary, making it difficult to get a Return on Investment (ROI).
Essentially, the idea was that I’d be able to go into any room at any time and ask them [HR managers] to hand over some random content… and we could put that content in the cloud and out would come what would resemble a course.
“We can’t help train everyone on how to build an incredible digital learning experience, BUT we can intelligently train an algorithm to build it for us. Essentially, the idea was that I’d be able to go into any room at any time and ask them [HR managers] to hand over some random content… and we could put that content in the cloud and out would come what would resemble a course. What we did was that we looked at the DNA of a course and we isolated 4 separate parts that we could address.”
Sahra-Josephine goes on to explain the DNA of an online course, what the four components are, and how they can be intelligently automated: The first part is an algorithm that catalogues all of the knowledge on the platform, and helps automatically assign knowledge tags to learner profiles. An example of Sahra-Josephine’s profile on the CanopyLAB platform can be viewed below.
“The ability to autotag, to use Natural Language Processing to catalog all of the information and accurately assign tags to people is enormously important. We built that first algorithm and we called it the A1: Auto-tagging algorithm. It tags every piece of information that comes in and then it summarizes all of the key learning in 5-7 tags.”
“The second part that had to be built to make the idea of a self-building course a reality, was exercise recommendations. On the CanopyLAB platform, we value helping people gain new competencies not just knowledge. That means, that learners are given the opportunity to engage in a series of different exercises to gain both knowledge and skills. But it turned out that there was a really big difference in supply and demand. Learners want different exercises than the one HR or teachers enable. That is something that the second automation feature addresses.”
Our A2: Recommender gives a series of recommendations for what exercises should be combined with the learning material.
There are still other parts of the course DNA for the team to address in order to actually enable the self-building course. One was the framing or introduction texts that come before learners are presented with the materials in a course. How close is that to being automated? Sahra-Josephine reflects:
“We’ve advanced so far with Natural Language Processing Summation… we are going to (in BETA) launch our NLP algorithm that can also auto-generate the text in the course in the first quarter of 2020.”
Someone has to build the exercises you might argue, and you’d be right! It’s the most difficult part, and the only part CanopyLAB is still keeping a secret.:
“So what is left is the exercises… The final part of what we are going to automate and we are still keeping it a secret how we are going to do just that.What we do see for most of our clients to be able to scale their digital learning whether it is in a refugee camp or in a consulting firm, these features will really help them.”
Just like with a self-driving car, you still have to actually be in the driver’s seat. It will be like that with the self-building course as well. You will be asked to be present and give some sort of hint as to where we are going. Instead of typing in the address on a GPS, you supply the content – for now.