Critical reflection post – generative AI for personalized learning via MOOC

I view massive open online courses with some skepticism, particularly when the source is a for-profit enterprise. Though the platform we chose has affiliations with credible post-secondary organizations and has been around for more than a decade, their intention is to be profitable and earn revenue through fees. The more learners who enroll in courses that have fees, the more money the company makes. My skepticism is rooted in questions around how much rigor is applied to evaluating student performance, and the temptation to be overly generous about prerequisites to enroll in a course to increase the number of students who are eligible.

The term ‘diploma mill’ is one that has come to mind when thinking of online learning. Whether that is a fair term to use in this case is beyond the scope of this blog post; I am merely reflecting on points that have been raised in our readings related to credential inflation, the awarding of digital badges, and the still-held belief that face-to-face learning is more credible than education delivered online.

Although the COVID-19 pandemic shifted our threshold of acceptance for how we work, learn, communicate, receive services, information, and conduct business, what should not change in all of these transactions is a desire for quality, critical thinking, and user-centred design. After reading George Veletsianos’ keynote remarks from the 2021 Congress, I thought about the learning event we are analysing, and indeed, where is the application of the four Es of effectiveness, efficiency, engagement, and equity? Proponents of platforms like Coursera can argue that MOOCs bring a measure of efficiency (meeting learning goals with a minimal expenditure of resources); however, what is the arbiter of other Es such as engagement and equity?

After experiencing this course intended for teaching personalized learning strategies, my view on this particular mode of delivering learning is still mixed. While I can see a utility and practicality of MOOCs in delivering skills training (particularly as I am a believer in life-long learning and upskilling throughout one’s career), I am influenced by Veletsianos’ remarks about being vigilant about the presence (or lack thereof) of the four Es in educational technology in all forms. This is especially salient if we are looking now at online courses which purport to instruct teachers on how to use generative AI to develop personalized learning strategies.

What are the criteria for these courses? Who reviews the curriculum for standards, ethics (a fifth E), veracity, accessibility, and sustainability? MOOCs may have expanded the reach of course content beyond the halls of a bricks and mortar school, but we cannot confound this with accessibility.

Reference:

Veletsianos, George. 2021, May 31. OTESSA (Congress) Keynote – effectiveness, efficiency, engagement. Where’s equity? https://www.veletsianos.com/2021/05/31/otessa-2021-congress-keynote-effectiveness-efficiency-engagement-wheres-equity/

Bias, assumptions, and stereotypes with AI and personalized learning

Our team was asked to select a technology and learning event to focus on for our critical inquiry analysis. We landed on the use of artificial intelligence in developing personalized learning strategies. The learning event we chose was a Coursera course called “Innovative Learning with ChatGPT”. The course description states that learners who enroll in this course will learn how to brainstorm lesson plans that integrate learner interests and needs, and how to personalize and customize educational materials for individual students.

In relation to this broader topic, for my own area of critical inquiry I will be looking into the issue of bias, assumptions, and stereotypes when using AI to develop personalized learning. Despite the encouraging overtones of courses such as the Coursera course our team found, one only has to do a simple Google Scholar search to see that personalized learning and AI continue to be the subject of critical reflection and inquiry – and for good reason. The limitations and shortcomings of data generated from language models such as ChatGPT can be numerous, including erroneous information, toxicity and bias, and manipulation of ideas (Hua and Jiang, 2023).

If we take these issues into account when thinking of developing personalized learning strategies, some alarming gaps are apparent. Take this text sample from the Coursera course, where the online instructor says about ChatGPT: “It’s not really thinking like humans do. It’s just really good at remembering all that information that it’s seen before” (Coursera, 2023). Who is the arbiter of the accuracy or impartiality of what data ChatGPT has seen? How can we be sure that the recommendations it will provide an instructor are in fact the best ones to use for a class full of nine-year old kids? ChatGPT doesn’t know anything about these children, except their age. What if the children are EAL (English as an Additional Language) learners? What if they are neurodiverse? While there is much that we have learned about what ChatGPT can do, it is nevertheless important to shine a light on what generative AI omits in developing personalized learning strategies. It is this with this lens that I intend to focus my research and reflection during this course.

References:

Hua, S., Jin, S., & Jiang, S. (2023). The Limitations and Ethical Considerations of ChatGPT. Data Intelligence, 1-38.

White, Jules. (n.d.). Innovative Learning with ChatGPT. Coursera. https://www.coursera.org/learn/chatgpt-innovative-teaching#reviews