This post was co-authored with Joan Oladunjoye

In recent years, the rapid advancement of AI technologies has reignited debates about the role of media in education. From generative AI models to AI-powered assistants, educational technology is often considered game-changing. The real question is, are these tools genuinely revolutionary, or are they merely delivering instructional content more efficiently? To explore this, we revisit the debate of Richard E. Clark and Robert B. Kozma in 1994, applying their perspectives to two contemporary examples of techno-deterministic thinking. We aim to understand how Clark and Kozma might respond to these claims and what their views could mean for today’s educational landscape.

In the original debate between Clark and Kozma, Clark asserted that media are mere vehicles for delivering instruction, having no direct influence on learning outcomes. He argued that what matters is the instructional method, not the medium itself. For instance, switching from textbooks to video lectures or AI tools would not necessarily change the learning outcomes; the instruction and pedagogy design makes the difference—not the medium (Clark, 1994, p. 26). Conversely, Kozma (1994, p. 18) contended that different media offer unique affordances that can enhance learning experiences. For example, a video provides visual cues that text cannot, and AI tools can offer real-time feedback, potentially reshaping how students learn.

Considering these perspectives, we examined two current examples reflecting the ongoing hype around educational technology.

Chat-GPT-4
The first example is from the article “Using GPT-4 to Improve Learning in Brazil”. In this article, OpenAI claims that GPT-4, a generative AI model, revolutionizes learning in Brazil by providing personalized tutoring at scale (OpenAI, n.d.). The article highlights the AI’s ability to adapt to individual learners’ needs, offering explanations, feedback, and tailored learning materials. According to OpenAI, this innovation is poised to improve learning outcomes dramatically.
We can speculate on their responses after reviewing Clark and Kozma’s arguments. Clark would likely assert, “It’s the method, not the medium.” He would argue that despite the enthusiasm surrounding GPT-4, the underlying instructional design will ultimately determine whether students benefit. For Clark, GPT-4 is merely another tool parallel to a textbook or a video (Clark, 1994, pp. 21-22). If the pedagogical approach remains unchanged, swapping GPT-4 would not significantly impact outcomes. Clark would caution against assuming that AI leads to better learning outcomes, emphasizing instead the importance of evaluating teaching strategies (Clark, 1994, p. 29).

Kozma sees GPT-4 as a transformative tool that showcases how media can influence learning. He might argue that GPT—4’s real—time adaptability—offering personalized feedback and tailored content—introduces new affordances that traditional methods cannot match (Kozma, 1994, p. 11). AI has the potential to transform learning by making it more interactive, engaging, and responsive to students’ needs (Kozma, 1994, p. 12).

Microsoft Copilot
Our second example is from the article “Microsoft Copilot for Education.” Microsoft’s Copilot, an AI-powered assistant, is integrated into various software programs to assist learners by generating documents, summarizing information, and providing real-time guidance during tasks. The article portrays Copilot as a “game-changer” for student productivity and creativity, asserting that this technology will significantly enhance learning outcomes (Microsoft, 2024).

Clark would likely respond by suggesting that “the hype is overblown.” He would express skepticism about Copilot being a “game-changer,” arguing that its mere presence does not necessarily lead to better learning. Like other technology, Copilot is simply a medium for delivering content (Clark, 1994, p. 26). If educators continue using traditional teaching methods such as lectures, worksheets, and tests, Copilot will be another way to support those methods. The tool is secondary to Clark’s instructional design and pedagogical approach.

Kozma’s argument supports the idea that different media, including AI tools, can enhance learning by providing unique affordances. He would likely argue that tools like Copilot promote creativity rather than limit it. According to Kozma, AI assistants offer new opportunities for learning by automating mechanical tasks (such as document formatting or information summarization), which allows students to focus on higher-order cognitive functions like creativity, analysis, and critical thinking. Kozma would see AI tools as enabling, not constraining, creativity by freeing students from routine tasks and providing interactive, real-time feedback that enhances the learning experience (Kozma, 1994, p. 13).

Both examples highlight a recurring theme in educational technology: techno-determinism, the belief that technology alone can drive educational transformation. Kozma might cautiously endorse this view, while Clark would vigorously critique it. As educators and technologists, it is crucial to remain mindful of overstating the impact of media on learning. Whether using AI tools, digital platforms, or traditional textbooks, these tools are only as effective as the instructional design behind them.

In conclusion, the Clark-Kozma debate remains highly relevant as educational technology continues to evolve with innovations like AI. Understanding the limitations and affordances of media helps educators make informed decisions about integrating technology into their teaching. While AI tools like GPT-4 and Microsoft Copilot offer exciting possibilities, they will not improve learning independently. Effective learning requires thoughtful instructional design, not just cutting-edge technology. By revisiting the Clark-Kozma debate, we hope to encourage critical thinking about the role of media in education and to prompt a more nuanced consideration of the claims made by educational technology advocates. As the debate continues, educators must balance embracing new technologies and staying grounded in sound pedagogical practices.

References
Clark, R. E. (1994). Media will never influence learning. Educational Technology Research and Development, 42(2), 21-29.

Kozma, R. B. (1994). Will media influence learning: Reframing the debate. Educational Technology Research and Development, 42(2), 7-19.

Microsoft. (2024, February 7). Delivering Copilot for everyone. Microsoft. https://blogs.microsoft.com/blog/2024/02/07/delivering-copilot-for-everyone/

OpenAI. (n.d.). Using GPT-4 to improve learning in Brazil. OpenAI. https://openai.com/index/arco-education/