Diminish costs

 

Design of educational content, ChatGPT and open educational resources: between competition and articulation ?

 

---

Auteur

Matthieu Cisel is an assistant professor at CY Cergy Paris University (France). His research lies at the intersection of educational sciences and computer science, with a focus on learning analytics and open education. Initially, his work centered on MOOCs, but he is now more interested in how teachers and trainers interact with Creative Commons licenses and, more broadly, how they share their teaching resources.

---

 

A quick glance at current events is enough to see that French universities are facing financial difficulties. The situation is unlikely to improve given the widespread budget cuts announced in an effort to restore public finances. This creates an ideal context to explore solutions aimed at introducing a form of economic rationalization into various aspects of academic work, particularly in the design of educational resources. Open education offers potential solutions in this regard. The topic is even more relevant today, as the rapid rise of generative AI (e.g., ChatGPT, DeepSeek) in conjunction with the use of open educational resources (OER) is likely to profoundly impact how course content is produced (Bozkurt, 2023). However, before delving into this issue, we must first clarify what we mean by “economic rationalization”.

This term should not be understood as an endorsement of a capitalist vision of higher education. On the contrary, the question is how to continue ensuring a high-quality public service despite constant or even reduced resources. One of the most obvious strategies is to pool the efforts of educators who, individually, create similar educational materials when synergies could save time for all. This is one of the core principles of open education. In this spirit, during the COVID-19 crisis, the FUN MOOC platform introduced FUN Ressources, a selection of openly accessible course materials (Massou, 2021). From the emblematic yet now somewhat outdated MIT OpenCourseWare to the OpenStax textbook repository, countless other examples could be cited.

At first glance, this is more about saving time than money, as universities do not pay faculty members for the time spent designing resources—only for their direct teaching hours. However, the time saved can be invested elsewhere, such as in research. From the students’ perspective, using open textbooks leads to significant savings (Wiley et al., 2012), especially considering the high prices set by academic publishers. This issue is particularly pressing in the United States, where student debt is the norm.

Returning to the question of time saving, this factor frequently appears in studies examining the mechanisms that encourage educators to adopt OER. However, my personal experience as a data science instructor has led me to question whether there is a form of competition, or at least a tension, between reusing resources created by others and generating content de novo with AI tools like ChatGPT. When the primary motivation for using OER platforms is to speed up content creation, individuals will leverage any available tool, often giving AI the advantage.

What teacher has not felt frustrated when failing to quickly find the right content for their slides or assignments, whether on OER sites or elsewhere on the web? Like many, I now turn to ChatGPT to fill in the gaps: generating LaTeX-formatted equations for statistical models, producing graphs to illustrate concepts, and more. Of course, the output is not always perfect, and I still frequently cross-check AI-generated content against reference sources. On occasion, when skipping this step, I have left minor errors in a slide—only to have students catch them in real time, reversing the usual classroom dynamic and putting me in the awkward position of having overlooked an AI-generated mistake.

That said, technology is evolving rapidly, and within my field, ChatGPT’s errors are becoming increasingly rare. Given the significant time constraints that all educators face, the principle of least effort often prevails when creating new content. This raises a recurring question: Should I rely entirely on OER platforms, use ChatGPT as the primary tool and trust its output, or use AI while rigorously verifying everything afterward? My decision-making process is always the same: given a constant quality level—since sacrificing quality is out of the question—which approach is the fastest?

Currently, the scientific community struggles to provide definitive answers to this question. The pace of technological change is so rapid that by the time a research project is initiated and its findings are published, they may already be obsolete. Moreover, the situation varies significantly across disciplines. In a survey of over a thousand French faculty members (Cisel, 2023), I found major differences in resource-sharing practices—for instance, computer scientists tend to be highly open, whereas law professors are generally less inclined toward open education. Extending this logic, we can assume there are differences across disciplines: AI-generated content is often of high quality in data science; however, given ChatGPT’s tendency to generate hallucinations, its usefulness for educators requiring precise references or book excerpts remains questionable.

Considering the rapid evolution of generative AI, research cannot yet provide definitive answers or broad generalizations across higher education. However, this should not deter researchers from tackling these issues. The key question now is: who has the determination and resources to conduct rigorous experiments—stopwatch in hand—to systematically analyze how generative AI and OER can be effectively combined to optimize the time spent by educators in designing teaching materials?

 

References

• Bozkurt, A. (2023). Generative AI, synthetic contents, open educational resources (OER), and open educational practices (OEP): A new front in the openness landscape. Open Praxis, 15(3), 178-184.

• Cisel, M. (2023). D’une discipline académique à l’autre, une approche contrastée de la diffusion et de l’appropriation des ressources éducatives libres au sein de l’enseignement supérieur. Distances et médiations des savoirs. (44).

• Massou, L. (2021). Usage pédagogique des ressources éducatives libres : quelles tensions entre ouverture et didactisation des ressources numériques ? Alsic. Apprentissage des Langues et Systèmes d’Information et de Communication. 24(2). https://doi.org/10.4000/alsic.5670
  DOI : 10.4000/alsic.5670

• Wiley, D., Hilton III, J. L., Ellington, S., & Hall, T. (2012). A preliminary examination of the cost savings and learning impacts of using open textbooks in middle and high school science classes. International Review of Research in Open and Distributed Learning, 13(3), 262-276.