Three Ways We’ve Used AI in Instructional Design
Recently, clients and others in my network have been asking if EnVision uses AI—particularly ChatGPT-style tools—and in what way. AI, in some form, has been around for a while. If you access your device using facial recognition or you’ve taken Netflix’s suggestion for your next video, you’ve experienced AI. In the L&D world, AI can be found in many of the popular elearning development platforms and tools, such as Articulate products and Vyond.
Some of my ID colleagues get upset about this: Isn’t AI taking jobs away from instructional designers? “No,” is my response. “It’s a tool we can—and need to learn how—to use in order to remain relevant.”
Here are some ways my team has started using ChatGPT, with due caution.
Creating a set of learning objectives and course outline
“Write learning objectives and an outline for a 1-day course on interviewing skills for managers.” This was my initial prompt in ChatGPT. It returned well-constructed learning objectives and a decent generic outline. But generic courses are available to anyone off-the-shelf.
Our clients typically come to us when they seek a customized learning solution. What if they need the course to focus on hiring people with specific skills or experiences, or have a particular culture they would like integrated into the course? What if the learners cannot leave their workplace for more than one hour at a time? We need to take many factors into consideration when designing and building out a course.
So, while ChatGPT might give me a head start, I do not use the output as is. I often need to iterate on my ChatGPT prompts, refining my request until it gives me something useful to my situation. The new skill of prompt-writing is critical to ensuring targeted output. It is one of the skills our team has been working on.
Drafting scenarios or video scripts
In addition to writing learning objectives, we have used ChatGPT to create draft scenarios and video scripts. Again, since our courses are customized to our clients’ specific needs, we must craft each prompt carefully. This often requires iterating.
Here’s an example of a prompt: “Write a script between a manager and employee where the employee needs coaching on speaking kindly to a customer.”
I found the response to be too generic, so I added: “ … Write the script in an emergency response feeding context.”
My output was better, but I wanted to ensure it incorporated the client’s feedback model. I added: “ … When the manager provides feedback, have them ask the employee first for their ideas.”
Next, I wanted to shorten the scenario, so I prompted: “Same idea, but keep it to under 1 minute of speaking time.”
You get the idea—iterate away until it’s no longer useful to do so. Then make your own edits.
While ChatGPT-type tools will not replace instructional designers, the tool can jumpstart components of our work. AI helps us work more efficiently and keeps us highly relevant.
Translating a course to another language
While there are dedicated translation AI tools (such as SmartCat), we can use ChatGPT to help translate short segments to another language. It is important to note, however, that this is simply a starting point.
We still require someone fluent in the translated language and familiar with the course content to verify the course’s terminology and preserve its original intent. For short translations, AI might be an efficient approach. Again, each situation is different and requires conversations with the client.
Summary and Cautions
While AI can aid us in developing our courses more quickly, it does not replace the critical analysis and practical application we as instructional designers perform. I look at it as a tool we need to learn, one that will help speed up parts of our process. At the same time, we must be thoughtful about how we use it.
No doubt, AI tools like ChatGPT will grow in effectiveness and popularity. We will continue to learn as we go—with excitement of the new technology balanced with a healthy dose of caution.
I’d like to close with three tips to consider before deciding to use ChatGPT:
- Carefully craft prompts—an art in itself—and then review the results critically and iterate.
- Consider context and whether the response you get aligns with what you need. You may need to refine your prompt. Or refine what’s useful and disregard the rest.
- Refrain from copy / pasting your company’s (or client’s) confidential content into a public tool like ChatGPT, as that could violate non-disclosure agreements.