AI Assist not AI Addict

The past few years various courses have mushroomed online – thanks to AI. All it takes is a prompt and you have your personal genie ready with a course. However, if you remember your fairytales well, you would recall that genies have their own set of conditions and wishes often come true with a catch, so one has to be sure what one wishes for. AI, for example, can generate a course in minutes, but that doesn’t mean you should publish it.

Creating an e-learning course used to be long drawn-out process with numerous checks in between to ensure that the final product was an effective learning solution. Someone had to write the content. An instructional designer had to structure it. A visual designer created the screens. A developer built the interactions. Someone recorded or sourced the voiceover. Then came testing, accessibility checks, revisions, localisation and publishing.

And then came AI – changing all of that. Give an AI tool a document and a few instructions, and you can get a course outline, learning objectives, screen text, images, knowledge checks and narration ideas in minutes. Authoring platforms are increasingly building these capabilities directly into the workflow. Articulate, for example, now offers AI-assisted course drafting and a growing set of AI capabilities inside Storyline and Articulate Rise.

Does this mean that the days of traditional e-learning development are over? Is instructional design no longer a relevant skill? On the contrary. It is much more relevant now. AI has eased the creation part. You want a course created; AI will assist you in the creation. The challenge lies in identifying whether the course created is in line with the requirement. Does it flow the way it should? Have the learning outcomes been addressed? Thus making instructional design all the more relevant.

The challenge start beyond the First Draft

AI has changed the economics of content creation.

A training team can take a policy document, product manual or presentation and quickly turn it into a structured learning draft. Text can be rewritten for different audiences. Images can be generated. Questions can be created. Existing content can be summarised and reorganised.

It is extremely useful to have that kind of roadmap to start with. What is important to understand is that the draft is not the answer, but a means to achieving the right answer.

A course can be grammatically perfect, visually attractive and technically functional and still fail as learning. It may check all the boxes but still fail to deliver learning outcomes.

Imagine a compliance course containing 30 screens of information.

AI can help create those screens. It can order information in the manner desired, but an efficient learning scenario would translate that information into a realistic scenario and prompt the learner to make the decision at the right time.

This requires understanding the learner, the context, the consequences of the decision and the behaviour the organisation actually wants to change – something that instructional design takes into account effectively.

Content Is Not the Same as Learning

Generating content is a piece of cake now. Ask a question. Get the information. Simple. AI has made it so. But you need to ask yourself if generating more and more content guarantees learning. Yes, you can have a huge library of content, but does it translate to learning? Not really. In fact, the opposite can happen.

The learner may become a passive consumer of information while the AI does the difficult thinking for them, effectively doing away with the learning objectives.

The OECD’s 2026 Digital Education Outlook highlights this concern: generative AI can improve performance on a task without necessarily producing corresponding learning gains when cognitive work is simply outsourced to the technology. The report argues for AI use guided by clear pedagogical intent rather than using AI merely to make tasks easier.

The Learning Conundrum

On one hand, AI is intended to make learning easier, while on the other, learning needs to offer challenges that can be overcome by the learner to grasp the given concept.

Sometimes the learner needs to struggle with a problem.

Sometimes they need to make the wrong decision.

Sometimes they need to retrieve information instead of being shown the answer.

Sometimes they need to explain why they chose something.

And sometimes they need to try again.

Those moments are not inefficiencies in the learning experience.

They are the learning experience.

The Role of the Instructional Designer Is Changing

It is more about management than creation. AI assisted development needs to be channelised into an efficient learning delivery.

The traditional production model looked something like:

Content → Storyboard → Design → Development → Testing

The emerging model looks more like:

Business problem → Learning strategy → AI-assisted creation → Human design → Interaction → Assessment → Measurement

The instructional designer moves further upstream.

Instead of spending most of the time creating every piece of content manually, they can spend more time asking better questions:

• What should the learner actually be able to do?
• What decisions will they need to make?
• Where are they most likely to make mistakes?
• What should they practise rather than simply read?
• What evidence will demonstrate that learning has occurred?
• What should happen when they get something wrong?
• What should change after the training?

AI can accelerate production. But instructional design makes the product relevant.

Stepping up Storyline

This is where tools such as Articulate Storyline become particularly interesting.

AI can help with the production layer. Storyline can still provide the control needed to create experiences that aren’t simply pages of generated content.

A policy course, for example, could become:

Read the policy → choose an action → see the consequence → receive feedback → try again

A sales course could become:

Enter a customer conversation → identify the customer’s concern → choose a response → see how the conversation develops

A safety course could become:

Observe a workplace → identify risks → make decisions → receive feedback

These experiences require more than generating information.

They require interaction design, scenario design, feedback design and instructional judgement.

Interestingly, Articulate’s own recent Storyline updates are moving in both directions: AI is being used to speed up production tasks while Storyline continues to provide extensive control over interactions, animation and custom learning experiences.

That is probably the direction the industry will settle into.

AI handles more of the repetitive work. Humans make more of the meaningful decisions.

Videos with an edge

AI has made its presence felt in all kinds of content creation. Of late a very obvious and popular use of AI has been seen across video production. Scripts can be drafted, voices generated, scenes visualised and translations produced far more quickly than before.

Tools such as Vyond make it possible to create animated training content without building every visual element from scratch.

The danger, again, is the use of AI blindly, with no focus on the learning outcome. Temptation of creating more and more videos may overshadow the reason for creating them –

If creating a five-minute training video is easy, why not create a ten-minute one?

Because the objective isn’t to produce ten minutes of content.

The objective is to create the right learning experience.

A short scenario that forces a learner to make a decision may be more valuable than ten minutes of polished explanation.

Production may have changed and evolved, but it cannot happen at the cost of compromising learning principles.

What Should AI Actually Do?

A useful way to think about AI in eLearning is to divide the work into three layers.

1. AI should accelerate

Tasks such as:

• First-draft writing
• Content summarisation
• Question generation
• Image creation
• Translation
• Voice generation
• Accessibility support
• Formatting
• Content tagging
• Production assistance

These are areas where AI can remove significant manual effort.

2. Humans should design

Humans should remain deeply involved in:

• Learning objectives
• Instructional strategy
• Scenario design
• Difficulty
• Feedback
• Assessment strategy
• Learner motivation
• Context
• Emotional relevance
• Business alignment

These decisions determine whether a course actually teaches something.

3. Technology should measure

The final layer is increasingly important.

A modern learning experience shouldn’t end with:

Course completed.

It should help answer:

• What did the learner understand?
• Where did they struggle?
• What knowledge gaps remain?
• What should they do next?

This is where assessment, analytics and AI can start connecting the learning experience with actual outcomes.

Identifying AI Advantage

Speed is no longer a differentiator for e-learning production. AI has more than adequately equipped developers to produce volumes at high speed. The real differentiator is the learning design. Is it relevant? Does it lead to the desired learning outcomes?

AI-Assisted, Not AI-Designed

The future of eLearning probably won’t be completely human or completely AI-generated. It will be collaborative.

AI will increasingly become part of the production environment—helping create drafts, visuals, assessments, translations, interactions and variations.

But humans will remain responsible for the most important question:

What should the learner experience, and why?

That distinction matters.

Because a course isn’t successful because it was generated in five minutes.

It is successful because, after five minutes, five hours or five weeks of development, the learner can do something they couldn’t do before.

And AI should help us reach that standard, not work towards lowering it.

Organisational Takeaway

The organisations getting the most value from AI in eLearning will probably not be those that simply generate the most content.

They will be the ones that rethink their learning development process:

• Use AI to reduce production effort.
• Use instructional design to create meaningful learning.
• Use interaction to make learners think.
• Use assessment to test understanding.
• Use analytics to identify gaps.
• Use human expertise to decide what happens next.

And those who understand this difference between being “AI Addict” and having “AI Assist” are the ones who will be future learning leaders.