The Evolution of Artificial Intelligence Pedagogy Moving Beyond the Tool Metaphor to Institutional and Medium Based Frameworks

The rapid integration of generative artificial intelligence into the academic landscape has prompted a significant shift in pedagogical discourse, moving away from simple adoption toward a more nuanced, critical understanding of the technology. Educators and scholars are increasingly challenging the pervasive "AI as a tool" metaphor, arguing that such a simplification fails to account for the systemic, infrastructural, and often unpredictable nature of large language models (LLMs) and other AI systems. Leading this intellectual push is Maha Bali, a professor of practice at the Center for Learning and Teaching at the American University in Cairo, who posits that artificial intelligence should be understood not as a tool, but as a "medium-institution." This shift in nomenclature carries profound implications for how higher education institutions approach digital literacy, student agency, and the preservation of human-centric learning environments.

The Limitations of the Tool Metaphor in Educational Contexts

For decades, technology in education has been framed through the lens of utility. Laptops, calculators, and learning management systems (LMS) are frequently described as tools—static objects that perform specific tasks under the direct control of a human operator. Maha Bali’s critique of this metaphor centers on the inherent predictability of traditional tools. In a recent analysis, Bali utilizes the example of a hammer: when an individual uses a hammer to drive a nail into a wall, the physical laws governing the interaction ensure that the nail moves in the direction of the force applied. The tool is passive, and the outcome is deterministic.

Artificial intelligence, however, does not share these characteristics. Generative AI models are probabilistic, not deterministic. When a user "hammers" a prompt into an AI, the output is not guaranteed to follow a straight path. Due to the "black box" nature of neural networks, the same input can yield different results, often containing hallucinations, biases, or unexpected deviations. By framing AI as a tool, institutions may inadvertently suggest a level of control and reliability that does not exist, thereby obscuring the ethical and technical complexities that users must navigate.

Chronology of the AI Pedagogical Shift: 2022 to Present

The transition from viewing AI as a peripheral utility to an institutional medium has occurred with unprecedented speed. The timeline of this shift highlights the reactive and then proactive nature of the academic community:

• November 2022 – January 2023: The Panic Phase. Following the public release of ChatGPT, the initial reaction in higher education was characterized by fear regarding academic integrity. Many institutions focused on detection software and outright bans, treating AI as a "cheat tool" rather than a transformative technology.
• February 2023 – August 2023: The Integration Phase. As it became clear that AI could not be effectively banned, the discourse shifted toward "AI literacy." This phase saw the rise of syllabus statements and workshops aimed at teaching students how to use AI for brainstorming, outlining, and coding.
• September 2023 – Early 2024: The Critical Reflection Phase. Scholars began to question the long-term implications of AI dependency. Concepts such as "techno-skepticism" and "critical AI pedagogy" gained traction. This period saw the emergence of frameworks like Taz Daniels’ Faculty Critical Engagement with AI Pyramid.
• Mid-2024 – Present: The Infrastructural Phase. The current discourse, championed by figures like Bali and Abi Awomosu, views AI as an inescapable infrastructure. The focus has moved from "how to use it" to "how to exist within it" while maintaining human agency.

AI as a Medium-Institution and the Invisible Hand

Drawing on the work of writer Abi Awomosu, Bali explores the idea that AI represents the "next industrial revolution." However, unlike previous revolutions where the primary concern was the physical displacement of labor, the current shift involves the mediation of thought and creativity. Awomosu’s analysis suggests that Gen Z students are not necessarily "booing" the technology itself, but rather the "invisible hand" that directs it—the corporate interests, data-mining practices, and lack of transparency that define the current AI landscape.

When AI is defined as a "medium," it is understood as an environment that surrounds the user, similar to how Marshall McLuhan famously stated, "The medium is the message." When it is defined as an "institution," it is recognized as a system with its own sets of rules, power dynamics, and social pressures. In this context, individual abstention becomes difficult, if not impossible. Bali argues that society is already "saturated with AI infrastructure," meaning that even those who do not use ChatGPT are still subject to AI-mediated systems in banking, healthcare, social media, and education.

Supporting Data on AI Adoption and Sentiment

Recent data supports the notion that AI has become an unavoidable infrastructure in academia. According to a 2023 EDUCAUSE QuickPoll, 54% of higher education respondents reported that their institutions were in the process of developing formal AI policies, yet only 11% had completed them. This gap indicates a lag between the technological "medium" and the institutional "governance."

Furthermore, a study by Tyton Partners in late 2023 found that nearly 50% of college students were using generative AI at least once a week, compared to only 22% of faculty members. This "usage gap" underscores the urgency of Bali’s call for literacy. If students are operating within an AI-mediated environment while faculty remain on the periphery, the ability of the institution to provide critical guidance is severely compromised.

The sentiment among students is also evolving. While early 2023 surveys showed high levels of excitement, more recent data from the Digital Education Council indicates a growing concern among students regarding the "homogenization of thought" and the potential loss of critical thinking skills. This aligns with Bali’s argument that literacy must serve as a form of resistance.

The Faculty Critical Engagement with AI Pyramid

To navigate this complex landscape, educators have developed frameworks to categorize different levels of AI involvement. Taz Daniels’ "Faculty Critical Engagement with AI Pyramid" is a notable example. This framework acknowledges that meaningful engagement does not require a "one-size-fits-all" approach. The pyramid allows for a spectrum of responses:

1. Thoughtful Abstention: Choosing not to use AI based on ethical concerns regarding data privacy, environmental impact, or pedagogical philosophy. This is viewed not as "being behind the times," but as a valid ethical stance.
2. Critical Awareness: Understanding how AI works and its limitations without necessarily integrating it into the classroom.
3. Controlled Integration: Using AI for specific, high-friction tasks where the human-AI interaction is closely monitored.
4. Full Transformative Use: Redesigning curricula around the capabilities and critiques of AI.

The benefit of this framework is that it validates "thoughtful abstention" as a necessary contribution to the academic discourse. It rejects the binary of "techno-pessimism" versus "techno-optimism," favoring a position of "literacy and sovereignty."

The Importance of Friction in Learning

A central theme in the critique of AI as a tool is the removal of "friction." In a pedagogical context, friction refers to the difficulty, effort, and cognitive struggle required to master a subject. Proponents of critical pedagogy argue that learning happens in the "spaces of friction."

In a debate involving scholars such as Jon Ippolito, Jeremy Douglass, and Marc Watkins, the concern was raised that AI serves as a "frictionless" interface that prioritizes speed and efficiency over depth and retention. If a student uses AI to summarize a complex text, they remove the friction of reading and interpretation. While this may be efficient, it may also be "hollowing out" the learning process.

Bali’s inclusion of the Indigo Girls’ lyrics "Hammer and a Nail" serves as a metaphorical reminder of the value of manual, "high-friction" work. The lyrics—"Learn how to use my hands / Not just my head / I think myself into jail"—resonate with the idea that over-reliance on abstract, mediated systems can lead to a form of intellectual confinement. The "refuge" of true understanding, the song suggests, comes from "tending the earth" rather than just posing for a "thoughtful pose."

Broader Impact and Implications for Higher Education

The shift toward viewing AI as a medium-institution has several long-term implications for the future of education:

1. Redefining Digital Literacy:
Literacy can no longer be limited to "how-to" guides for software. It must include an understanding of the political economy of AI, the environmental costs of data centers, and the algorithmic biases that reinforce social inequalities. As Bali states, "Literacy is resistance." It is the capacity to engage deliberately with a medium that one is already inside.

2. Preserving Human Agency:
The goal of education in an AI-saturated world shifts from "avoiding contact" to "preserving agency under conditions of contact." Institutions must empower students to decide when to use AI and when to rely on their own cognitive processes, ensuring they are not "enchanted into dependency" or "shamed into secret use."

3. Ethical Labor Practices:
By recognizing AI as an institution, universities are forced to reckon with the labor practices behind the technology. This includes the underpaid data laborers in the Global South who "clean" the datasets that power LLMs. Critical pedagogy demands that these ethical considerations be part of the curriculum.

4. The Sovereignty of the Learner:
Ultimately, the discourse led by Bali and her peers argues for the sovereignty of the learner. In a world where the "invisible hand" of AI infrastructure seeks to predict and direct human thought, the role of the educator is to provide the tools—and the friction—necessary for students to remain the masters of their own intellectual development.

As AI continues to evolve from a novelty into a foundational layer of global infrastructure, the academic community’s move away from the "tool" metaphor represents a vital maturation of the field. By treating AI as a medium and an institution, educators can better prepare students for a reality where technology is not just something they use, but something they must learn to inhabit critically and consciously.