Artificial Intelligence (AI) is everywhere in eLearning today, transforming how we create, deliver, and personalise eLearning experiences. From AI-powered content recommendations to interactive avatars guiding learners through complex topics, artificial intelligence is rapidly shaping the future of digital learning.
Clients started asking us to add AI avatars in to their eLearning courses in early 2023. Some loved them, some didn’t.
The clients that were on board with using AI avatars in eLearning appreciated the benefits. The cost is far lower than hiring actors. The time needed to create a talking head video is also significantly less. Mistakes in pronunciation are fixed within minutes, without the need for additional studio time. Include the fact that adding video can make eLearning more engaging, and you’re on to a winner.
Another major benefit is that by using AI we can quickly adapt content into different languages, voices, and styles, making it easier to reach diverse learners around the world.
Feedback from those that didn’t care so much for the AI avatars centred around the fact that they found the voices to be robotic, and the movements to be wooden. Certain words can sound strange too, which detracts from the overall believability of the whole avatar-as-a-human thing.
I understand where the feedback comes from, but I’m in the camp who find that the benefits far outweigh the negatives.
Okay, so the lip-synching isn’t perfect, and some of the facial expressions can become a little repetitive on longer videos, but having the ability to generate talking head video in minutes rather than days is a major benefit when timelines and budgets are tight.
Take a look at our sample below, what do you think?
Challenges of using AI Avatars in eLearning
It isn’t all plain-sailing though. As its use increases, so too do the ethical questions that surround AI. In turn, so do the number of challenges faced by eLearning developers.
At the beginning of our AI journey, we worked with a Pharmaceutical Industry client to develop an Objection Handling course. The script that the client provided was based on a sales representative holding discussions with various customers within a healthcare environment (Physicians, Payers etc). In the discussions, each healthcare professional presented a barrier to using the product the sales rep was promoting, each time referencing the client’s organisation name, for example, “I don’t feel as though we get the best prices from Company X”. In early 2023 the AI avatar service we used generated the videos without issue.
Fast-forward to late 2024 and we were asked to update the videos. Not with a new script, but simply to take advantage of the improved avatars with less robotic voices, and less wooden movements. This time however, the AI service rejected our videos, on the basis we had violated the guidelines around ‘Inauthentic Behaviour: Impersonation’.
Unfortunately the guidelines weren’t clear enough to help us resolve the issue, and nor was the online help. So after a quick internal discussion, it was suggested that we try removing the name of the client’s organisation from the script, and hey presto, problem solved! The trouble is, the script had been approved by the client. As anyone who works in the Pharma industry will know, changes to approved content are a very time-consuming thing …
In absence of any clear guidance from the AI vendor, I can only conclude that a fear of litigation was a contributing factor in previously approved videos being blocked this time around, so beware when writing your scripts!
Barriers to using AI avatars in eLearning?
The majority of our work comes from the Pharmaceutical, MedTech, Healthcare and Life Science industries. Again, in case you don’t work in any of these areas, the sensitivity of client data is a BIG deal.
One of the barriers we’ve encountered is that clients simply can’t allow any propretary information to be entered in to the AI systems. With Machine Learning (ML) being closely linked with many AI systems, clients aren’t able to take the risk that their content might end up being used to train an AI with their sensitive data.
I would suggest that how we navigate this is beyond the remit of your average eLearning developer. In fact, this is very definitely an issue that must be handled by our clients’ legal teams before we spend even a second developing scripts.
Are AI avatars a viable soultion in eLearning?
Despite the cautionary tale above, yes, I think AI avatars do have a place in the eLearning developer’s toolkit. Okay, so there’ll be a handful of legacy videos that we’ve created previously that will need to be modified during the usual update cycle, but now that we know about the updated guidelines, we can work within them (until they evolve again I suppose).
Actually, it’s reassuring to know that in a world where AI appears to be taking over, that there is actually something in place to reduce the number of deepfakes out there.
It might even be a saving grace for human acting and voiceover talent too as they look to where their next gig might come from. There are a lot of good people in these industries who will find themselves short of work as a result of Generative AI, so if we find ourselves facing insurmountable barriers to using AI for video or voiceover, try reaching out to a human who’ll have no such issues giving you exactly what you need.
Related eLearning Posts
How to continue numbered lists in Articulate Rise
16s video explaining how to continue numbered lists after inserting a new block in Articulate Rise 360.
Creative Data Visualisation: Ballon d’Or
Now and then I like to have a bit of fun and create interactive infographics.
Drag and Drop to Enlarge Images Screencast
Screencast explaining how to use Articulate Storyline 2 to create a Drag and Drop to Enlarge Images interaction.
David began his eLearning career in 2000 after earning qualifications in Graphic Design and then Newspaper, Magazine and Infographic Design.
During his time working in the eLearning industry David has gained extensive experience designing and developing eLearning content for clients based in the UK, Canada and the USA. To date, David has created eLearning courses covering a broad spectrum of topics, ranging from Pharmaceutical, Healthcare, Retail and Sport through to Pet Nutrition, Marketing, Sales and Tourism.
He is passionate about creating well-designed and purposeful eLearning solutions. David is a current Articulate E-Learning Heroes Super Hero.
