Generative AI
Pioneering AI-Driven Instructional Design in Small College Settings
For institutions that lack the budget or staff expertise to utilize instructional design principles in online course development, generative AI may offer a way forward.
One does not need to be a higher ed insider to notice the increasing number of colleges integrating instructional design (ID) principles into their course design workflows. Historically a staple of the corporate Learning & Design (L&D) space, ID refers to the discipline that focuses on elements of course quality extending beyond the subject matter itself, to questions surrounding "how" that material is presented, through the students' perspective.
On a broader level, it is the discipline that bridges the gap between on-ground and online learning by examining the ways in which students engage with course content. The benefits of ID are well-documented in the literature: Students in ID-enhanced courses experience improved persistence, retention, and achievement of learning objectives.
Unfortunately, smaller colleges — arguably the institutions whose students are likely to benefit the most from ID enhancements — frequently find themselves excluded from authentically engaging in the ID arena due to tight budgets, limited faculty online course design expertise, and the lack of ID-specific staff roles. Despite this, recent developments in generative AI may offer these institutions a low-cost, tactical avenue to compete with more established players.
At Lackawanna College — a smaller institution of higher education in Northeast Pennsylvania — our team has been piloting the use of generative AI in course development workflows. Through our pilot, we have been able to develop impressive course content that is arguably on par with many larger online institutions — and may represent a sustainable ID path forward for similar colleges.
Defining the Rationale and Challenges
Our journey started by asking self-reflective questions, the most important of which was: "What is the highest yield and lowest cost ID intervention that we can implement from a student success standpoint?" This lens brought us back to the basics of examining "how," on a text level, the subject matter itself was being presented to students. We realized that many of the trendier ID-interventions of today, such as adaptive learning and gamification — while absolutely important to student success, and ones we will revisit in the future — may need to take a back seat to the most critical ingredient of all in effective course design: the language being used within the course.
We realized that our students resonated the most with courses written in a professional, yet conversational style. Instead of terse, third-person language — similar to reading a textbook — we realized that students were more likely to engage with lessons and assignment directions where they felt like the instructor was truly speaking "to" them, akin to being an audience member to a captivating speaker.
We also realized the importance of context cues within components of a learning module. For example, in an on-ground class, an instructor often naturally gesticulates for emphasis, and we realized that capturing these "off the cuff" gestures in written form should be a critical area of focus to make our online courses even more authentic and relevant for students.
At the same time, we also recognized that the lack of dedicated editor and content writer roles — aside from the subject-matter expert (SME) — would make it challenging to implement these language-based initiatives across the full suite of online courses, let alone in a single online course. These challenges were reinforced by time-on-task calculations performed by our Instructional Technologist. These hurdles certainly weren't unique to our institution; to the contrary, they are frequently the norm at smaller colleges and universities. Despite these obstacles, our eLearning team leaned into a solutions-focused approach.
Adapting AI to Instructional Design
The reality is that a news article is being published almost daily about higher ed faculty using ChatGPT to draft lesson plans, create interactive learning materials, and personalize educational content for different learning styles. At the same time, notable case study applications of these tools on the ID front are limited, or at best, restricted to theorical models or more niche applications (gamification, personalized learning, predictive feedback, and real-time assessment). We wanted to pilot the use of these tools on a more practical front — as a "great equalizer" — to determine if we could achieve and generate course quality exceeding the bounds of an institution's existing ID scale. In other words, rather than us singularly looking at AI as a niche strategy for enhancing specific elements of an online course, we wanted to pilot a few AI-driven ID interventions, and lay the groundwork for a broader conversation on how we continue to offer a high educational value proposition to our students.
At the same time, this wasn't an endeavor we entered into lightly. We were well aware of the concerns associated with AI use, including issues of bias and fairness, information accuracy, intellectual property, and pedagogy. There are valid philosophical critiques of AI, many of which I agree with, centering on what it means to be a human being, and the inherent value (and definition of) receiving a "quality education." And yet, as we've seen throughout modern history, this pattern of concern accompanying the advent of new technology is nothing new. In an ironic twist, it's this ingrained hypervigilance — the inherent fear of change — that defines us as human beings. It is a trait that is both a blessing and a vice; one we must accept and voluntarily make the decision to set aside, when warranted.
We knew that inaction was not the right approach; those players often find themselves playing catch-up years down the line, ultimately losing their innovative flavor. We also knew that we wanted to keep our students' best interests at the forefront. That's why we opted to take a responsible approach by piloting these tools on a small scale first, before attempting any broad-based initiatives in the future.
Tangible Applications into the Instructional Design Process
As we began our pilot, we found ourselves returning to our central theme—the importance of language and authenticity in better reaching our students. The strategies explained below, while unique to our institution, may provide inspiration to others.
The Introduction to a Module. Rarely, if ever, does an on-ground instructor begin a class session by jumping headfirst into the course material. More often than not, even a novice instructor will begin an on-ground class with pleasantries, housekeeping items, and perhaps some informal musing on the connections between the upcoming lecture and the content covered in the previous class. In the online realm, this aspect often does not come intuitively for SMEs designing the course. Not only does it require a formal understanding of curriculum design, but it also requires an understanding of generational differences in how students process tone in written text. We realized through our pilots that generative AI can effectively output these conversational, human-like lead-in phrases at scale, especially if detailed prompts are used. We specify tone, demeanor, and request a "professional, yet conversational," writing style in the prompt. We have also had success by specifying the module's learning objectives and key lesson content from the module.
The Relevance of Subject-Matter Content. We have also found AI to be highly effective at expanding upon, or clarifying, existing subject-matter content. In the ID industry, course materials often undergo multiple levels of revision, which is an iterative process that can be challenging to implement in a small college setting. However, we have found that with the responsible use of AI, we can enhance the existing subject matter in ways that maintain its integrity, while also generating additional examples, rephrasing dense verbiage, and providing related topics for exploration. Aside from creating more engaging course material, this approach reframes the broader conversation from purely designing curricula, to designing curricula that targets the demographics of a given institution. Something as simple as prompting AI to generate a metaphor for a complex scientific principle that resonates with an 18-24-year-old demographic may be just what is needed to improve a student's connection to the course.
Discussion Questions and Assignments. In the on-ground classroom, even the broadest style of discussion question — as an extreme example, the question of "What is life?" — can easily evolve into dynamic, nuanced discourse as ideas and questions get bounced around in real time. In the asynchronous online environment, those same questions often fall flat. These types of simple, linear questions also tend to be the types that students may feel tempted to answer using generative AI. As a solution to this, we're finding that AI can be proactively used to revamp discussion questions and create prompts that tap into higher order thinking. This not only drives better student-to-student communication, but also mitigates the risk of academic dishonesty.
Module Summaries and "Looking Ahead" Sections. We are also finding that the end of a learning module is often just as important as its beginning. In on-ground settings, instructors often verbally summarize key points or set the stage for the next session at the end of class. In online courses, the end of a given section often ends abruptly, due to the simple fact that online courses tend to be structured as compartmentalized learning modules. Through the use of generative AI, we have been able to craft concise, coherent summaries that not only wrap up the core learning from the module, but also give students a preview into what will be explored in the upcoming module. Not only does this help with overall course congruency, but we have heard from students that this helps them to see the larger context and value of the course.
Key Takeaways
There's no denying that our exploration into AI in instructional design is still in its infancy and that the more we explore, the more questions we seem to have. While we are excited at the long-term possibilities these approaches hold for our program's growth, we are also acutely aware that we will make errors, and will need to grow and learn from those mistakes. Although the ID work at Lackawanna College is far removed from the cutting-edge developments in machine learning and neural networks at R1 institutions, what we are currently experimenting with is a microcosm of what will be transpiring across the higher ed ID industry in years to come.
In the future, I hope to see a more equitable technology landscape in higher ed, irrespective of institutional size. As an administrator at a smaller college, I believe that taking these small steps in implementing AI into the instructional design process is one of many examples of micro-innovations that bridge the very best of the intimate, small college experience with the world class learning experiences of larger schools. Through that lens, AI shouldn't just be viewed as a tool for creating better learning experiences, but as a true catalyst for driving a more inclusive and equitable future for students at large.
