Note: A great many of these ideas are half-baked and off the cuff.

The internet is saturated with companies constantly vying for your attention. The learning tech industry is no different. As with most sectors, learning tech is using predictive analytics to understand what you want so that you stick with their product. This has led to quite a few discussions between Tim and I about how we structure Experience. In short: what should be automated and what shouldn’t be. As we head into the future there is a growing expectation from everyone that tasks will be automated. Algorithm, Artificial Intelligence (AI) and Machine Learning (ML) have become popular buzzwords. They power most modern apps in some form or another. One of the questions that we asked was: “Should we be using these buzzwords to serve learning content?” The conclusion that we came to was: “No.”

Perhaps it sounds a bit odd in a world where you can purchase a Kohler Numi 2.0 toilet, powered by Amazon’s Alexa (I’m not making this up). Far from being Luddites, we just believe we need to give proper consideration to automation. We’re asking, whether we should be trusting computers to make learning decisions for us. If we do let them make learning decisions, what does that mean about the things that we’re learning.

Netflixification

The Netflix logo.

On my trip to the Learning Tech conference this year, there were a host of companies that were promoting LXPs (Learning Experience Platforms). Some people that visited the LXP stands were asking the question: “what is an LXP?”. The most common answer: “The Netflix of learning”.

So what is the “Netflix of Learning”?

Put simply: An LXP both provides learning content on-demand to its users and automatically makes suggestions to a user about what they should learn about next. Think LinkedIn Learning.

It’s part of the current trend of the netflixifying everything. And it’s growing fast. Several arguments are made for netflixification:

  • It’s cheap
  • It pushes lots of content
  • It keeps people coming back, again and again
  • It can be easily sold as a cheap subscription service
  • It’s very easy to scale

All of these points make it a very appealing to L&D departments. Buy an off the shelf platform, get your cohort setup up with logins, and everything they need to know will be pushed to them. The problem that an LXP claims to solve is discovery. An LXP will recommend content based on:

  • What you have viewed before
  • What you’ve told it your interests are
  • What it “thinks” is contextually relevant

In principle, this sounds like a good idea. Why spend the time money and effort on curating courses for learners when the computer can do it for you?

Why This Could be a Bad Thing

As Netflix’s platform has matured and stabilised, it is becoming apparent that it might not be as clever as we think. Login into your Netflix account and check it out yourself.

Your dashboard is likely to be structured in the following way:

My current Netflix dashboard.
  1. Headline: The latest Netflix movie or series release
  2. My List: Content that you, yourself, have actively marked as being of interest
  3. Because you added something else to your list: Movies and series that tenuously match something else you’ve watched
  4. Continue watching: Stuff you’ve already started watching and therefore shown an interest in
  5. Trending now: Stuff that other people are watching
  6. Subheadline: Usually a series that they’re trying hard to push
  7. Everything else below is a random assortment of increasingly narrow genres.

A lot is often made of the Netflix algorithms pushing content to people as an explanation for why Netflix is so popular. The reality though, is that Netflix is pretty lazy when it comes to figuring out what their audience want to watch. Consider the first 5 rows from the description above:

  1. Headline: Something Netflix wants me to watch
  2. My List: Things I have told Netflix that I want to watch
  3. Because I added something else to my list: Matching content based on similar metadata.
  4. Continue watching: Things that Netflix already knows that I like
  5. Trending now: Things that everyone else seems to like.

Call me a cynic, however this doesn’t come across as being particularly clever. It does give the impression that it is though. Who wouldn’t be somewhat impressed if, when logging onto a website, they immediately saw things that they liked? This is the illusion of Netflix. This is also true of netflixified platforms. They show you things that you like, because you’ve normally told it exactly what it is you like. This isn’t “discovery”; it’s pigeonholing.

A pigeon.

This is why we need to be extremely cautious in learning tech. We run the risk of autonomous, self-curating platforms suggesting that the users “discover” the same things repeatedly. Even if you stop someone seeing the same course again and again, they see the same topics. This pushes them into increasingly narrow schools of study and thought. Great if you’re pursuing and PhD, perhaps, not so much for continuing professional development (CPD). 

CPD is usually about building a broader base of transferrable career skills. If you want to become a warehouse manager, there is no single skill you can develop to become one; you need a skillset. Let’s think about the skills that a warehouse manager might need:

  • Logistics
  • Supply chain management
  • Transport management
  • Customer service skills
  • Maths
  • Business management skills
  • Critical and lateral thinking
  • Stakeholder management, etc.
Oversimplified diagram suggesting the skills required to be a warehouse manager.

Trying to achieve these skills through an LXP might be quite interesting. How would you go about discovering these skills if you didn’t know what they were? The LXP suggests that based on your profile, it would push these out to you. But what if the above skills are wrong? Full disclosure: I am not, nor have I ever been a warehouse manager. I am not qualified to suggest skills that warehouse manager should have. I have made what I consider to be a best guess as to some skills I might expect them to have. It is a guess, but a somewhat informed one. The LXP on the other hand is not “guessing”. It’s not even thinking. Because it doesn’t think, it doesn’t care. This is perhaps the most critical part of a educator/learner relationship. Would you not prefer a person (who was able to understand the full context of your goals) to make learning suggestions?

The LXP Learner Journey

Let’s imagine that you, as a learner, have decided to learn more about finance and have searched the LXP’s catalogue. 

A netflixified learner dashboard might look like this:

  1. Headline: We can ignore this for the sake of this argument
  2. My List: A list of finance learning content selected by me
  3. Because you added finance to your list: Similar finance, money, accounting, etc. courses
  4. Continue: Continue your current finance course or a course you’ve previously completed
  5. Trending now: Courses that other people are doing.
The Sisyphean learner.

As you can imagine, the learning experience provided to the learner in this type of scenario isn’t great. This may well prove to be a poor example, but I think you get my point.

Not only does it not suggest anything other than that which the learner has already expressed an interest, the suggestions that it can make are without total context. One solution that has been put forward by IBM (as part of it’s Watson enhanced LXP) is to create a catalogue of roles and append skills to that role. The taxonomical structure would end up looking a bit like this:

  • Warehouse manager
  • Skill 1
  • Skill 2
  • Skill 3
  • Skill 4
  • Finance Manager
  • Skill 1
  • Skill 3
  • Skill 3
  • Skill 4
  • etc.

It then uses this structure to align the pushed content to the user’s role (and therefore assumes their desired progression path). This can be modified by the user so that the suggestions are more relevant.

The problem with this approach, however, is that not all roles and job titles mean the same thing. Even within very similar companies. Beyond that, new job roles are being created all the time and similar roles may be referred to by very different names. You end up having to make so many exceptions to the rule, that the idea of a rule itself becomes silly.

So what is the solution: The Learning Concierge

The Learning Concierge

Humans are achieving increasingly more with increasingly less effort through technological advances, like automation. With this we have to be brave enough to realise that somethings just aren’t meant to be automated. Learning is one of them. Learning should be about personal development, achievement, and progression. You are an autonomous agent, that decides what you should be learning. There is always scope for help though. Removing automation doesn’t mean people being left to their own devices. We need to rethink the role of the teacher/mentor.

“How can I help you learn more stuff good?”

The Learning Concierge is not a new concept. Jane Hart wrote a great piece on the concept in 2013. The concept, however, does not seem to have taken off. 

The idea of the learning concierge, is effectively someone that you can talk to who can help direct your learning. They are a personal service that recognises that there is no “one-size-fits-all” approach to learning. It is staffed by real people, and is ideally made up of people familiar with the area that the learner is interested. You can speak to them, they understand you, and the broader context of your goals.

This is, in essence, what netflixification is attempting to do. It is trying to replace the role of a curator (or concierge) to allow their business model to work at scale. This may be perfectly fine for entertainment purposes, but when a person’s development rests on their learning outcomes, is this what we should be doing?

Experience as a Platform for Learning Concierges

To get the most out of a learning experience, it must be customised to the learner’s unique needs. At the moment, this isn’t possible to do using AI in a meaningful way. What we can do, however, is create learning communities within an organisation. This means allowing people with the right skills and experiences to act as learning concierges to others. You allow them to share their skills and experiences, suggest courses, reading lists, videos, podcasts, etc. to those who want to achieve similar things. Content is king. 

This already happens pretty frequently, most just don’t recognise it as a learning concierge service. If you subscribe to the What Matters newsletter (or any other popular newsletter or podcast) you’re already doing this kind of learning. Influential people often aggregate and share their knowledge and experiences. The problem is though, that none of this learning is recorded even though it is legitimate.

An LRS enabled LMS can record learning experiences from anything. Literally.

Experience is built in a way that addresses this problem. By having a Learning Record Store (LRS) at the heart of the platform, we can record engagement with any content type. This means that all those videos, news articles, podcasts, etc. that you’ve learned from can be accounted for. It means that a concierge can see the things that people like and are engaging with and the things that people aren’t engaging with. There are two key benefits to this:

  1. The platform can host workspaces for people of similar development paths with content relevant to their needs
  2. It encourages individuals in an organisation to share their knowledge 

This, we believe, is the key advantage that an LRS enabled LMS will have over any LXP: The concierge will be able to suggest better learning content, resulting in better learner outcomes.

How the human brain learns in animated, pictorial format.

So what do you think? Are LXPs the future or do you still, like us, think people are preferable?

If you’re interested in seeing how Experience can be used as a Learning Concierge platform, head over to experience.nucleus.ac and try it for free!

Curtis Anderson

I've been a learning experience design and learning technology expert for the past 7 years. I've worked on some amazing projects in my career so far including the creation of OUP's Kerboodle VLE, the Foreign and Commonwealth Office's Diplomatic Academy, and the MoD's UKMFTS. In March of 2019 I joined forces with my business partner, Tim, to create our own company. We now spend our time finding solutions to some of the biggest problems facing digital learning.

Leave a Reply