I say this because while different roles in L&D (trainer/facilitator, producers, L&D technologist, and more) will be affected to varying degrees by the rapidly evolving technology, the role of instructional designer (ID) will see significant, permanent alteration by generative AI.
I feel confident in making this assertion about instructional design and generative AI because I have experience in this field: my corporate career began in 1999 as an instructional designer, and I did that work—or led people who did—for about eight years.
We created classroom materials for in-person and virtual instructor led training, as well as e-Learning modules that included simulations, video, interactive components, and more. We had an elaborate process for creating all of this learning content—both off-the-shelf and customized—for our customers on a very wide range of subjects, from technical training, to soft skills, compliance, and more.
What we didn’t have was a way to quickly go from nothing—a blank word document, empty PowerPoint deck, a non-existent e-Learning storyboard—to something that was half-way decent and ready for an initial review. That was always the hardest part of each project: over and over, starting with little or no content, and slowly putting together words, audio scripts, images, animations, videos, and more, depending on the output modalities required.
But all of that is changing—and fast. On the same level as the calculator, the personal computer, and the internet, generative AI is and will disrupt many, many jobs. To make this point with one example, lets follow step three in i4cp’s Strategic Workforce Planning process by deconstructing the job/role of instructional designer into some of the most common tasks involved, and considering for each how existing (or soon to exist) generative AI tools will disrupt, enhance, or augment that work. Consider these common ID tasks:
- Perform initial research on a topic. Traditional internet searches still have a place (e.g., to find well-composed articles by trusted and favored sources). But leveraging generative text tools like ChatGPT provides access to well-rounded, essentially crowd-sourced content, due to its having learned from a massive Large Language Model (LLM) that spans a huge swath of available global content. Some specific prompts that IDs likely find helpful in creating training materials would be requests for: key research findings on the subject; information on critical facts, concepts, principles, processes, and procedure; information on barriers, challenges, and unintended consequences to consider, etc. Generative AI tools can also be used to summarize lengthy material found via traditional searches, such as videos, articles, recording webinars, books, etc., allowing an ID to consider a broader set of inputs than ever before.
- Interview subject matter experts (SMEs). Some questions that IDs ask SMEs are standard, but if you don’t know much about the subject (after all, you aren’t the expert!), it can be hard to know the best questions to ask to get the most out of your limited access to their valuable time. An ID could use an appropriate generative AI tool with a prompt such as “Give me seven questions to ask a subject matter expert on topic X that will provide insights for a training course.” If an ID is struggling in their interaction with a SME (not an uncommon situation), generative AI could also come in handy to help generate more nuanced, follow-up questions to ask.
- Create an initial outline. With initial research and the output of SME interviews in hand, IDs next create a content outline to give themselves a path to follow as they create the training program materials. Without this, creating documents, slides, interactive e-Learning, assessments, simulations, etc., will quickly get out of hand. The right generative AI tool will provide IDs with one or more initial outlines to consider.
- Flesh out the written content. Whether text in a training manual or scripting for voiceover audio, IDs write a lot of words. Generative AI can turn paragraphs into bullets, and vice-versa. It can help source famous or inspirational quotes to support a point. It can help with bridging sentences to keep the training flow moving along, or craft “what’s in it for me” (WIIFM) statements to keep learners engaged.
- Edit and improve written content. Learning content should be engaging, but it’s easy for even the best IDs to fall into wording repetition traps or fail to effectively explain a key concept. Generative AI will help via prompts such as “Provide four examples that illustrate X,” “Reword the above three points to make them more relevant to an audience of engineers,” “Rewrite this paragraph to use simpler/more persuasive/more engaging language,” etc.
- Generate ideas for multimedia. Effective instructor-led training materials and self-paced e-Learning content leverage a wide range of slides, images, videos, animations, etc. It is sometimes a challenge to come up with good media, especially for more abstract concepts, principles, and processes. Leveraging media-generating AI tools can either spark ideas for the IDs and graphic designers on a project, or in some cases produce final media outputs. Consider the following prompts: “Provide me with three analogies for process X,” “Give me five images that show a leader influencing others, but in different contexts”, or “Create a series of PowerPoint slides, with appropriate imagery, from these three checklists.”
- Create quizzes/assessments from existing content. At some point, an ID will be asked to create quizzes, assessments, etc. Doing this is both a science and an art—but there is also a lot of tedium involved in the task. With the right generative AI tool, an ID’s training manual or other learning content could be entered with a request for say 15 knowledge test questions, in specific formats, following given parameters, etc.
- Improving the accessibility of learning content. Like writing assessments, another important but time-consuming task for IDs is ensuring that the learning materials are accessible to everyone. This can be as simple as adding alt-tag captions for each image in an e-Learning module, but this can involve dozens or hundreds of images in a lengthy program. Letting a generative AI tool loose on this task will free the ID’s time, requiring only a review of the output for accuracy, and adjustments where needed.
- Create performance support and retention materials. A typical formal blended learning program might involve a mix of instructor-led sessions, self-paced modules, videos, reading, etc. But the best programs also provide performance support tools for use when needed, back on the job, as well as micro-learning content to aid in learning retention over time. The best IDs will keep such additional content needs in mind while designing a program, but creating it at the end of the process can be time-consuming and keep them from moving on to their next project quickly (or worse, these important outputs get cut from the process for that very reason.) Generative AI can be used to quickly create learning nudges in various formats to help combat the forgetting curve (the human tendency to not retain much of what we initially learn), or to create checklists, job aids, etc., for performance support on the job.
I could go on—depending on the project requirements, there are many more tasks that instructional designers routinely do. Some, such as performing a learning needs analysis or developing the initial learning objectives for a program, might benefit from generative AI less than some of the tasks above.
But you get the point—much of what instructional designers do is generating learning content, so it’s not surprising that the wide range of generative AI tools that we are all witnessing coming alive in 2023 will disrupt the profession in a profound way.
Recommendations for L&D Professionals
Notice that nowhere in the above did I say that generative AI will eliminate the role of instructional designer. Rather, to borrow a phrase I’ve heard applied to other jobs, instructional designers who use AI will replace instructional designers who do not use AI.
Generative AI will forever change the remit of the instructional designer; augmenting the role and making it more productive—immensely more productive. Once increased efficiency is achieved, some organizations might choose to keep the quantity and quality of their training material constant and simply employ fewer IDs. But I predict far more organizations will choose to leverage greatly improved ID productivity and either produce more training material faster, better training material, or both.
So where to begin? Here are four recommended steps for L&D leaders and practitioners to quickly and smartly reap the benefits that generative AI is already bringing:
- Get clear on the basics. Not all AI is generative AI. L&D has been using AI for many years and in many ways, such as Netflix-like algorithms in LMS/LXP systems that make personalized recommendations for each learner, or scripted chatbot tools that were the next generation of FAQ documents for performance support. Those were and continue to be valuable, but generative AI is different in being far more responsive and by literally generating new content.
- Consider the impact of generative AI for your organization’s L&D strategy. You are likely to find that once generative AI becomes widespread, there will be less need for formal training courses as such. In short, this moment is an inflection point in the long-desired shift away from creating large libraries of training courses towards a greater reliance on performance support and learning in the flow of work.
- Upskill, upskill, upskill. What one upskilling need was not on anyone’s radar in late 2022, but is now, or will be soon? Promptcraft. That concept/skill will surely be added to the Oxford dictionary soon, but L&D professionals—IDs and others—need to learn the art and science of creating prompts and seed texts to get the most value from the generative AI tools they will soon be using. This will likely entail learning some basics and then practicing—a lot. Additional upskilling will vary, such as how to use specific generative AI tools and how to review, evaluate, and attribute the outputs.
- Work with IT, compliance, and others to set guidelines for the use of generative AI. There are already countless articles that detail the many issues that can arise from generative AI. These are real: false information from ChatGPT and similar tools, data leakage (sensitive data being shared externally) when using some tools, bias in how the AI was trained, humans ceding too many decisions to AI, and more. All of these are already creating a certain level of FUD: fear, uncertainty, and doubt. Guidelines are needed to mitigate any risks and cut through the FUD fog so that the equally real benefits of generative AI can be realized.
- Pilot the use of each generative AI tool. No one tool is going to augment all of the ID tasks listed above, let alone tasks performed by other L&D professionals and beyond. Select appropriate projects to pilot the use of each tool being considered, working with key stakeholders to stay within organizational guidelines.