AI and the e-learning industry

The source of inspiration for AI

We know that in order to respond to our requests, AI feeds on all the content available on the web. The machine learns by analyzing tons of data, looking for trends among all the content posted online, sequences, and drawing inspiration from them to “create” the new content requested by the user. In other words, its food is the content of the web.

However, what we find on the web consists of free material, sometimes produced by people with no experience or expertise in e-learning production. Very often, these are the work of novices and, dare we say it,

• the pedagogy is poor
• and the results are far from convincing.

And for a very good reason: productions made by professionals in the field are not available to AI. If an e-learning producer creates high-quality training for a client, it is not to make it available free of charge to its competitors. This production will remain buried deep within the client's internal network and AI will never see it. Why would the client make this production available to everyone, especially their competitors?

This means that when it comes to e-learning, the machine has a very poor tutor!

The quality of the source material given to AI

For their part, the AI user mentioned above fed the machine a PowerPoint presentation summarizing all of its content. Their approach is transmissive and the teaching techniques are weak; the AI will therefore produce transmissive e-learning with minimal media interactivity.

And that's not even mentioning pedagogical interactivity! Because it's important to make the distinction: while media interactivity interests and stimulates, pedagogical interactivity makes you think. The two forms must work together to produce a quality product.

Evaluating the quality of the final result

Obviously, the final result will be hampered by these two factors:

• a machine that has learned its job poorly
• a source of content that is undoubtedly reliable, but transmissive in its approach.

The saddest thing is undoubtedly the fact that the AI user will be unable to make a critical judgment about its output, for three reasons.

1. Since it will find most of the knowledge from its original file, it will reassure the user and seem effective in their eyes.
2. Their judgment is likely to be biased by the time savings, which will certainly be impressive.
3. Having never been in contact with high-quality e-learning content themselves, they are unaware of the criteria for evaluating it.

This is a crucial point. Most customers are unable to recognize quality in e-learning because they have no reliable benchmarks to rely on. They cannot distinguish between an e-learning production that informs and one that trains! For novices, “informing is training,” when we know very well that learning is much more complex and demanding!

Healthy competition for e-learning producers?

According to figures collected in the industry by Chapman Alliance, it takes approximately 200 to 350 hours of work to produce one hour of e-learning. This varies greatly depending on the level of interactivity desired by the client. The richer the interactivity, the more the learner invests themselves, the more they learn, but the more complex it is to produce.

Let's imagine an hourly business cost of $100, which is certainly not exaggerated, and we end up with an outlay of $20,000 to $35,000 for a quality one-hour production. But if the client doesn't know how to evaluate quality, how can producers compete with AI, which will take 20 minutes, often for free?

I admit that it's not very encouraging when viewed from this angle. It will be difficult to keep high-quality e-learning production alive in such a context. There are still a few solutions.

• Raise customer awareness of the importance of choosing high-quality training if they want to see observable and lasting results.
• Help customers distinguish between “information activities” and “learning activities”;
• Teach AI to work properly.

In any case, there is work to be done if we don't want to delude ourselves into thinking that our participants are learning when all they are doing is clicking the “Next” button. A brain in streaming mode does not learn...