Skip to main content

So we have been tooled for the last few days in production with the AI Creator. 

And there are a few concerns, but before we get into it lets start with the positive notes:

  • these are exciting times
  • these tools are set for a trajectory to answer speed to product concerns
    • and if you havent been addressing some speed to product by adopting a few AI assisted generated workflows with content with your IDs? Then it is safe to respectfully say that you are starting to lag in todays market

Now that we have covered some of the positive notes, I would love to start a real conversation about these tools and beginning to adopt them.

I am on with the creator tools testing them out some during the evening of 7/1. It is a quieter time in the Eastern US. And I just tried building my first AI Presenter video.

And it’s output was really? Decent.

And so I got a first test and feel for this.

I have to admit though, I have alot of concerns about authoring with these tools just yet.

My top 5 concerns with using it as is today:

1 - Video has to support captioning today. It is not an option, it is a must. And certainly with a push of more of the ADA going into action as of recent? We are being charged as practitioners to support different modalities of training when we are generating media.

2- As of July 2025? There is some serious lag when it comes to generating content. We aren’t talking about measuring in seconds. We are talking in the count of minutes to generate even a ten second video.

3- The layout of the AI Video generator feels a little too simple. Video should support an easy way to wrap a bit of title/caption text around it (on top, to either side of it, on the bottom of it) without us needing to do a round trip to the other blocks of content. Something kind of like this graphic:

Proposed Layout choices. Consider that these should be the different layout blocks.

4- Just a white background? Could we add some flavor of a background for these folks? Like an image for a background or even a color?

5-What exactly is happening with my minutes of AI generation usage being tracked?

Is there a freemium model going on here? Whats the upcharge?

And that is not even tackling the challenges associated with turning on an AI content generator that “consumes”/stores our documents.

I want to call a spade a spade to say IMHO that we are what feels like at least a year out before I can let content creators / authors loose with these tools at my org. And that is where I figure - it would help to also hear from you.

Can you use these tools “as is” to generate modern elearning today? Did I miss a glaring concern? Are you using a tool that can do this better? Am I over reacting and we should just start using it right now to build content?

Thanks in advance for adding to this in a constructive meaningful way.

 

 

My struggle with AI is much the same as my struggle with using robot voices from text to speech applications - I cannot rely on delivery of what I consider a quality product.

The argument I hear so often is that it is sooo much faster but  that is not my experience. With text to speech - it is fast to upload your text and get some speech but the speech just sucks or you have to spend a ton of time finding ways to spell things differently just so you can get the words to come out the right way that it would have been faster to just read the script yourself.

Same with these AI tools - I cannot trust them to come up with accurate content regarding my industry/product. Often - I spend a great deal of time proofing and modding the content that I should have just done it myself from the get go.

Best use I have found so far is to generate some quizzes based on content I provide but there are a large number of bad questions or good questions with bad answer choices.

I don’t have time to mess around trying to find the right prompt to get what I want. It is faster to just make what I want.

​@gstager - as always thank you for your perspective.

I am at a point where I (we) have given into what we think is high quality for text to speech generated audio. Industry specifics  I get and there are tools that many of these use to maintain a persistent dictionary for you.

My consistent go to has been WellSaid Labs…but because of pricing and competitors stepping up in quality? We have recently started leveraging Elevenlabs (not very inventive names - eh?). 

I do think they are at a point where they dramatically improve our “speed to product” without invoking my own voice or paying for voice talent….but that is my humble opinion.
 

 

I have no issues recording my own voice - It sounds better and is usually a lot faster. I did make use of WellSaid one time on a project - it did have the best result of the others I tested at the time. The experience, overall, however, was not great.

I only used it since I needed a ton of different voices both male and female. For projects where I can just do it myself - that is what I do - At least I don’t have to spend time trying to make it sound like I know how to speak basic English.

​@gstager I feel your pain, unfortunately I don’t have a good “radio” voice at all so recording myself isn’t really an option, and it was getting too cumbersome to try to find people either in our department or others to read scripts. Of course the claim from leadership is there is “no budget” to hire a real VO artist so I was tasked with finding an AI solution. We tested Murf and were pretty happy with their tools and VO voice library (tons of languages too) but the cost was too much for this budget year, so instead we’ve signed on with Lectora’s Microbuilder; we got a prorate and discount since we’re already a Lectora customer. I feel like they have the same character set as Docebo’s Creator, which is interesting, but I’ve managed to get some good audio results out of it so far. It depends on who you choose; some voices definitely sound more natural than others. My feeling is over time, the issues with pronunciation will improve to where less effort will needed to spell things out just the right way (especially website addresses!). We’re just in a frustrating spot right now where the tech is emerging, we as instructional designers are being pushed to use it to boost efficiency, but there are still many areas where it’s still not as good as the “old” way of doing things.

Reply


Terms & ConditionsCookie settings