It has been a while since I have either worked with or interacted with George Siemens, Stephen Downes, and David Wiley. I can’t claim to have ever been that close to any of them, and a lot has probably changed in the years since I really followed any of their work. I work in the instructional design field, and they don’t, so unfollowing them wasn’t anything personal – it was just they aren’t in my field and now that I am mostly out of academia (thankfully), I decided to focus just on my work.
So I was quite surprised to see George Siemen make a very anti-instructional design comment in a recent post that says “I’m thinking course design should cost about zero.” Stephen Downes followed up with the statement of “My thought too.” I know that people often make quick statements without thinking through words, so I am hoping there will be clarification as to what they meant here. Hopefully they didn’t mean that my entire field of instructional design should go away (after all, “course design” is pretty much what we do), or that it should be replaced with Ai. I don’t think either one of them is that ignorant of what instructional designers do. But, since that statement is all I have, I will have to take what they said at face value and show why it is wrong either way.
The confusing part is that they are both pointing to a post by David Wiley detailing some updates on his OELM idea. I have some concerns with that model as well, but I will get to that in a minute. I first have to point out that Siemens and Downes both allude to OELM as example of why course design cost should be zero (even though Downes points more directly to OELM than Siemens does) – but Wiley’s post makes it clear that OELM will NOT make course design costs zero. It may even make them more than they currently are.
Take, for example, several quotes from Wiley’s post that allude to how institutions will have to pay someone to do the design:
“If you don’t like any of these the way you find them, you can revise or remix them before reusing them.”
“Here the instructor or instructional designer pre-selects combinations of topics and activities and creates links to these bundles and places them strategically throughout the course.” (Note: Wiley doesn’t fully think of “activities” in the same sense that instructional designers do, more on that later.)
“the designer or instructor does this alignment step manually by associating specific OER with a topic.”
“As currently conceived, OELMs supplement existing learning materials like textbooks and courseware – they don’t replace them. This part of the design is also strategically important. Instructors will not adopt OELMs in place of the textbook or courseware they already use because generative AI is too unfamiliar to them. But I’m developing what I think is a very solid strategy for talking about OLEMs in a way that will make it relatively easy for instructors to adopt OELMs as supplemental materials.”
“developing OELMs should include a collaborative process in which groups of faculty and students work together on the OER curation and adaptation, activity type selection, and activity specification / instruction writing.”
If you are looking to bring the cost of course design down to “zero,” OELMs are not the way to go.
Of course, there could also be the erroneous assertation that there will be no costs when using Ai. Even assuming that Ai companies will keep their tools free (which, come on – no way they will do that forever), there is still a cost in resources… a massive one for electricity and water that is not going to be solved any time soon. There will never, ever be a “zero cost” Ai solution – just maybe one where you can pretend it costs you zero dollars personally (and the rest of the world be damned).
Maybe Siemens and Downes are looking at just the beginning of Wiley’s post, and thinking that somewhere within the idea of separating content from presentation lies the secret sauce to making course design costs zero? If so, I hate to break it to them, but Wiley gets the idea of separating content from “presentation” wrong.
After making a comparison to CSS, Wiley makes this statement:
“If you think about learning activities as a way of “presenting” learning content, then you can start thinking about how to cleanly separate learning content from learning activities.”
I can hear the instructional designers out there groaning (I have already talked to a few that did so out loud). Activities are not a way of “presenting” content. Think of Bloom’s taxonomy action verb charts. Now, I know there is nothing scientific about how these charts are organized, but they contain hundreds of verbs of what you can do with learning activities – very few that would fall under the category of “presenting.” Of course, if you think of flash cards and quizzes as “activities,” then maybe all activities are presentations. Yes, there are learning activities (some very popular) that fall under the “presenting” category – but most don’t.
Then there is this misunderstanding about writing content:
“people typically write textbook chapters without thought for learning activities beyond reading and highlighting.”
I used to work for an educational publishing company, and I have also reviewed hundreds of textbooks as an ID. For at least several decades now, almost all textbooks come with sample activity sections at the end of each chapter, and typically entire ecosystems of activities in the form of supplemental books and online portals. Textbook companies have those activities mapped out first before most content writing starts. Even when individual professors write textbooks on their own, there is no way they separate that writing from the years of activities they have assigned as well. I would say it’s more true that it is difficult to write content without thinking of activities. Of course, it can be done – but who is going to shut off years of learning and possibly teaching experience when they write?
“However, it is wildly different to try to imagine creating learning activities without making any specific references to content.
While it is true that activities and content are quite different, it is actually easy to create activities without content. I have designed many courses with no content and all activities. There are many courses out there without textbooks, FYI. Content is important for many courses, of course. But not all.
But back to the comparisons to HTML and CSS. As someone that does a lot of coding (still) in HTML and CSS, I find Wiley’s use of these coding languages as an example of what OELM does with content (HTML) and activities (CSS) to be problematic. On the surface level, what he describes as the interaction between content and activities somewhat lines up with how HTML and CSS work in a finished website. But getting to that finished website? You have to constantly go back and forth between HTML and CSS, because changes in one effects the other and vice versa. This site is hosted in WordPress – where a lot of the HTML and CSS interaction is already completed. But I have still spent hours tweaking both the HTML and CSS at the same time because they are much more entwined than Wiley is implying here. The same applies to designing a course. When the course is finished, sure the content and activities (somewhat) work like Wiley describes. But while you are in the process of designing the course? Not at all.
Wiley then gives an example of a learning activity that just involves a quiz. Technically quizzes should be assessments and not activities, but many people conflate the two (including myself sometimes as well). However, there is a massive range of learning activities beyond quizzes or flash cards or debates, many of which would not work very well within the confines of an Ai-generated quiz “activity” that is basically customized feedback loops:
“After each answer, give me feedback on my answer and explain anything it seems like I don’t understand.”
The fact that most Ai companies have disclaimers that they are not giving you facts, but just “most likely responses based on the data they were trained on” (which, let’s fact it, is mostly garbage), would invalidate the pedagogy of this request. Why should we trust these systems to quiz students and give them feedback when there is the chance it could be incorrect or nonsense? Wiley even later touches on a good reason why you should avoid using Ai for this but doesn’t comment on the problem at all:
“The purpose of the content is to keep model responses accurate and reduce hallucination.”
Reduce hallucinations? Why throw learners into something that can “hallucinate” (I know, poor way to describe it, but it is what most people know it as) at all?!?!
Wiley also puts this instruction in his example:
“If I get distracted or try to change the topic of our conversation, politely but firmly refuse to talk about topics not contained within the <content> section.”
There is so much that is wrong with this statement, not the least of which is how students sometimes learn new information by connecting it to random other topics. That is outside the scope of this post, but don’t do things like this to learners. Remember that they are not robots that have to “stay on task” just to learn.
Wiley then goes on to imply there is something unique about OELM for remixing courses:
“Cleanly separating content from presentation (like HTML / CSS does) greatly increases the reusability of the learning content and the learning activities.”
Again, there is no clean separation between HTML and CSS, even in finished websites. That promise of CSS never fully came to fruition. But, myself and most instructional designers quite easily reuse content and activities in courses over and over again all the time – even in less instructor-focused courses where students determine what they study. I did it quite easily when I utilized self-mapped learning pathways in my dissertation work. Sure, some LMS systems like Canvas make this reusability much easier, but even clunky older Blackboard installs can still do it somewhat easily. Your “average instructor” has been able to “create, revise, remix, and share” content and activities “without any special technical knowledge” for decades. OELM is not unique in this.
The first example of how to use OELM as a prototype (labeled “exploratory”) that Wiley gives probably sounds familiar to readers of this blog (all 3 of you). Educators have been using tools like Twine and even SoftChalk to set up “choose your own adventure” type activities like these for quite a while. I even did an interactive presentation at OLC Innovate in 2019 about chatbots, Ai, and game theory that was basically the same thing but with the Ai that was available at the time and Twine holding it together. Whether you have to learn how to craft prompts or use a Twine interface (or whatever tool you use), it’s all about the same level of difficulty for the end user. Ai doesn’t make it better or faster, just different. However, using Ai for the whole system gives you less control than putting it together in something like Twine. Which some people prefer, but then the amount of time you have to spend fixing the problems you run into after the Ai gives you something is about double what it would be to just do it yourself in the first place (in my experience).
Wiley does touch on how there will be a need for different activities for different courses, and the selection of those activities will require knowledge of current research. But he is still not dealing with the fact that most learning activities sit outside the realm of “presenting,” and many courses utilize activities that are outside the reach of LLMs.
None of this is to say that OELMs are a bad idea, just that they are incredibly limited, and to think you are making a “system simple enough for average instructors to work with” misunderstands what the average instructor wants in their course. Clean separations of course content and learning activities are generally labeled as bad pedagogy in instructional design circles – and not really realistic in my experience with hundreds of different courses.
This is where the experience and expertise of instructional designers would come in. While instructional designers get mentioned only twice in Wiley’s post, they get left out at the most important parts, such as:
“Finally, developing OELMs should include a collaborative process in which groups of faculty and students work together on the OER curation and adaptation, activity type selection, and activity specification / instruction writing. Again, more about this in a future post.”
Faculty are famously not trained in how to design courses, and those that learn to work with instructional designers are often impressed by how well their courses turn out. Instructional designers know that the interaction between content and learning activities is too nuanced and complex to trust to Ai, and that learning activities are much, much more than just “presenting content.”
This leads me back to “what were Siemens and Downes thinking?” I’m really not sure. The article they use to support their point doesn’t support zero cost for course design, and it is only useful for certain types of courses. Maybe the answer lies in this quote from Siemens:
“Btw, how do you read papers now? I read, drop into Gemini for summary. Upload to NotebookLM for a short audio. Discuss with ChatGPT. It’s less about reading and more about interacting and engaging.”
Having used Gemini, NotebookLM, and ChatGPT – I can’t believe anyone would seriously do this. These tools get so much wrong in my experience. Not to mention that ChatGPT has massive bias, including a very pro-Ai bias that it rarely drops (even when you have to constantly point out it’s bad logic, like I did with this post). I wish I had captured the comically bad (but also concerning) conversation I had with a Character.ai bot the other day.
Since most deep instructional design knowledge wasn’t included in the training data for most LLMs, I’m guessing Gemini and NotebookLM gave a junk summary of Wiley’s article (because I have tried those tools as well, and while their summaries are usually accurate, they are always biased in favor of Ai and usually missing the things instructional designers would include in the summaries) and then ChatGPT gave it’s usual pro-Ai discussion. I could be wrong, but I still stand by my encouragements for people to turn off Ai and use their own brains instead. Or at least read Audrey Watters to balance things out.
Matt is currently an Instructional Designer in the private sector and a Part-Time Instructor at the University of Texas Rio Grande Valley. Previously he worked as a Learning Innovation Researcher with the UT Arlington LINK Research Lab. His work focuses on learning theory, Heutagogy, and learner agency. Matt holds a Ph.D. in Learning Technologies from the University of North Texas, a Master of Education in Educational Technology from UT Brownsville, and a Bachelors of Science in Education from Baylor University. His research interests include instructional design, learning pathways, sociocultural theory, heutagogy, virtual reality, and open networked learning. He has a background in instructional design and teaching at both the secondary and university levels and has been an active blogger and conference presenter. He also enjoys networking and collaborative efforts involving faculty, students, administration, and anyone involved in the education process.