How I Used AI to Turn a Document into a Course Storyboard & Prototype in <10 Mins
An experiment, demo & prompt for you to try
Hey everyone!
This week, I’ve had a great time working with the folks on the June cohort of my bootcamp to explore the question: does AI make it possible to write a high-quality course storyboard and build a working prototype in under ten minutes?
Here’s what we learned…
1. AI is Not an Instructional Designer
The first thing we learned very quickly here is that AI needs you - i.e. the quality of the prompt that you give to AI is critical. If you simply ask AI to “create me a course storyboard on X topic” it reproduces instructional design practices which are common rather than those which are optimal.
For example, one of ChatGPT’s “bad instructional design habits” is designing according to learning styles which, despite being debunked, have been around and spoken about for long enough to be prominent in AI’s training data.
So, the first key learning for us was: AI needs you. Specifically, it needs your instructional design expertise. The “garbage in, garbage out” AI phrase has become a bit of a cliche but it remains very true.
🥡 Key take away for instructional designers: as well as instructing AI what to do, tell it how to do it.
Example:
You are an expert instructional designer. Using the information attached, you will create a first draft of a course storyboard.
When creating the storyboard you must:
Apply active and scenario learning principles which enable the user to learn through experience rather than content consumption.
Use Gagné’s Nine Events of Instruction to structure the sequence of learning content, activities and feedback.
Never refer to or design using “learning styles”.
2. AI is Not a Subject Matter Expert
The next thing we learned was perhaps less of a surprise: AI is not a reliable subject matter expert.
Most generic AI tools like ChatGPT and Copilot are trained to be general assistants. This means that they will always try to help you and always give you an answer, even if it means it must hallucinate some or all of it in the process.
The result is that AI is often much more confident than it is competent; if you ask it a question, it will make a best guess and deliver that best guess with a level of confidence which can be very compelling - but also very incorrect.
So, the second key learning is that - once again - AI needs you. Specifically, it needs you to provide reliable, verified domain information as its knowledge base.
🥡 Key take away for instructional designers: as well as instructing AI what to do, and how to do it, define the resources it must refer to as its source data.
Example:
You are an expert instructional designer. Using the course outline and the compliance documents attached, you will create a first draft of a course storyboard.
When creating the storyboard you must:
Apply active and scenario learning principles which enable the user to learn through experience rather than content consumption.
Use Gagné’s Nine Events of Instruction to structure the sequence of learning content, activities and feedback.
Never refer to or design using “learning styles”.
Only refer to the documents provided.
Cover all content in the document provided.
3. AI is Now a Builder as Well as a Writer
One of the key affordances of "multi-modal" AI models like ChatGPT-4o and Claude Artifacts is that they can generate not only text but also various types of assets, including those that instructional designers commonly build.
The result is that if we have access to free AI models like ChatGPT-4o and Claude Artifacts we effectively have access to a new breed of rapid prototyping tools.
So, the third key learning in this project was that AI can help us no only to design but also build (and in the process reapidly test and iterate) learning experiences.
🥡 Key take away for instructional designers: we can work with AI to generate a well-formed storyboard and then turn that storyboard into a working prototype (or a v1 of it).
Example:
You are an expert instructional designer. Using the the compliance documents attached and the instructions below you will:
1. First, create a course storyboard.
2. Second, create a first draft of a working prototype of the storyboarded course design using HTML, JS & CSS formats.
When creating the storyboard you must:
Apply active and scenario learning principles which enable the user to learn through experience rather than content consumption.
Use Gagné’s Nine Events of Instruction to structure the sequence of learning content, activities and feedback.
Never refer to or design using “learning styles”.
Only refer to the documents provided.
Cover all content in the document provided.
When creating the prototype you must:
Write all course text
Create code for all interactive elements in HTML, JS & CSS
Insert placeholders with descriptions for all multimedia elements
As you’ll see in the demo below, with the right prompting we were able to turn a dry, dense compliance document into a well-designed course storyboard and a working prototype using Claude Artifacts in <10 mins 🤯.
Amid all of the excitement of seeing weeks of work turned into minutes of work, perhaps the biggest lesson we learned this week is AI Needs You.
In a world where AI is increasingly part of our day to day lives and work, there’s an irony that deep human expertise becomes arguably more critical than ever.
For instructional designers, the rise of AI will likely create a future where we spend less time on “functional” tasks like building content and more time honing a deep understanding of how humans learn and develop so that we can become great teachers, instructors and critics of our AI apprentices.
Happy innovating!
Phil 👋
If you want to get hands on and try experiments like this with me and a handpicked group of innovative learning designers and educators, apply for a place on my AI-Powered Learning Design Bootcamp.