Beyond Adoption: Navigating Institutional Change in the Age of AI

Author: Denis Avetisyan

Higher education is facing a new era of disruption, demanding a shift in change management models to address the unique challenges presented by rapidly evolving artificial intelligence.

This review proposes a framework for adapting institutional change theory in STEM education to accommodate generative AI as an ‘arrival technology’ that necessitates a focus on uncertainty and adaptation.

Existing models of institutional change in STEM higher education presume a period of established practice preceding scaled adoption, a logic challenged by the rapid arrival of generative AI. This paper, ‘A Framework for institutional change in the age of AI’, proposes a new framework reconceptualizing institutional adaptation under conditions of genuine uncertainty, identifying six dimensions-spanning tool characteristics and stakeholder roles-that necessitate a shift from adoption to ongoing inquiry. The framework emphasizes humble, localized approaches, pedagogical grounding over tool-specific solutions, and the repositioning of change agents as facilitators of collective exploration with students as partners. How can institutions proactively leverage this framework to navigate the evolving landscape of AI and foster meaningful educational transformation?

The Inevitable Erosion of Institutional Memory

The history of higher education is characterized by a constant cycle of reform and innovation, yet surprisingly few changes achieve lasting integration into institutional practice. While colleges and universities readily adopt new programs, pedagogical techniques, or administrative systems, many ultimately fade into obsolescence or are abandoned altogether. This phenomenon isn’t necessarily due to a lack of merit in the innovations themselves, but rather a complex interplay of factors including inadequate resource allocation, insufficient faculty buy-in, and a failure to address deeply ingrained cultural norms. Consequently, institutions often find themselves in a state of perpetual ‘pilot fatigue,’ repeatedly launching initiatives that, despite initial promise, struggle to scale and sustain impact, highlighting a critical need for more nuanced and holistic approaches to educational change.

Historically, efforts to reshape higher education institutions have frequently relied on directives originating from administrative leadership, a methodology that often overlooks the crucial role of faculty and the specific contexts of individual departments. This top-down approach, while seemingly efficient, can inadvertently stifle innovation by failing to account for the practical realities of teaching and research, or to harness the expertise of those most directly involved. Studies reveal that changes imposed without meaningful faculty input are less likely to be fully adopted or sustained, as they may not align with existing pedagogical practices or address locally identified needs. Consequently, institutions seeking genuine and lasting transformation are increasingly recognizing the importance of fostering a more collaborative and bottom-up model, one that empowers faculty to become active agents of change and tailors initiatives to the unique characteristics of their respective disciplines and student populations.

The emergence of “arrival technologies” – notably generative artificial intelligence – is disrupting traditional educational change protocols by fundamentally altering the pace and pattern of innovation adoption. Unlike previous technological integrations subjected to lengthy pilot programs and rigorous pedagogical assessments, these tools are being rapidly embraced – and, crucially, integrated – directly into learning environments, often bypassing established institutional review processes. This presents a predictable challenge, as the perceived benefits of AI-driven personalization and accessibility are weighed against anxieties surrounding academic integrity, data privacy, and the long-term impact on critical thinking skills. Consequently, educators and administrators find themselves in a reactive position, attempting to evaluate and adapt to technologies already in widespread use, rather than proactively shaping their implementation to align with established learning objectives and institutional values.

Building on Shifting Sands: Established Foundations for Change

Interactive Engagement (IE) instructional methods have consistently demonstrated positive impacts on student learning outcomes across diverse educational settings. These methods move beyond traditional lecture-based formats by actively involving students in the learning process through techniques such as think-pair-share, group problem-solving, and in-class demonstrations. Empirical studies indicate that students in IE classrooms exhibit significantly improved performance on concept inventories and standardized tests compared to those in passively-lectured courses. The effectiveness of IE is attributed to its emphasis on cognitive engagement, allowing students to construct their own understanding of concepts rather than simply memorizing information, and fostering a more inclusive learning environment through increased student participation.

The Diffusion of Innovations (DoI) theory, originating with Everett Rogers, describes how new ideas and practices are adopted within a social system. This process isn’t instantaneous; rather, individuals fall into categories based on their adoption timeline: innovators, early adopters, early majority, late majority, and laggards. Critical to successful diffusion is the identification and engagement of early adopters – individuals who are respected by their peers and willing to experiment with new approaches. Effective communication channels, both interpersonal and mass media, are also essential for creating awareness, building understanding, and ultimately driving adoption. The theory emphasizes that perceived attributes of the innovation – its advantages, compatibility, complexity, trialability, and observability – significantly influence the rate of adoption.

Departmental Action Teams (DATs) and the Science Education Initiative (SEI) exemplify successful, faculty-driven approaches to educational transformation. DATs, typically composed of faculty members within a specific department, facilitate localized change by identifying challenges, proposing solutions, and implementing evidence-based practices. The SEI, a broader initiative, similarly relies on faculty leadership to redesign curricula and teaching methods, often focusing on active learning techniques and incorporating data-driven assessment. Both models prioritize faculty ownership and collaboration, recognizing that sustainable change is more likely when those directly involved in teaching are actively engaged in the process of innovation and implementation. The success of these initiatives demonstrates that bottom-up, faculty-led efforts can effectively address pedagogical challenges and improve student learning outcomes.

Adapting IE Protocols for an Accelerating Entropy

Existing implementation and evaluation (IE) change models are insufficient for navigating the rapid and expansive integration of artificial intelligence. Traditional models often focus on adoption rates and overlook the accelerating pace of technological development and the broad scope of AI tool application. A revised framework must account for the non-linear rate of change characteristic of AI advancements, as well as the diverse contexts in which these tools are being deployed – ranging from narrowly defined tasks to systemic organizational shifts. This necessitates a move beyond simple adoption metrics to encompass ongoing adaptation, iterative refinement, and continuous evaluation of AI’s impact on existing processes and workflows.

The proposed framework for adapting instructional engineering (IE) change models moves beyond traditional adoption-focused approaches by utilizing six dimensions: pedagogical, technological, content, temporal, ethical, and affective. These dimensions address the multifaceted challenges presented by generative AI, acknowledging that successful integration requires consideration beyond simply implementing the technology. The pedagogical dimension focuses on shifts in teaching practices, while the technological dimension addresses infrastructure and tool selection. Content considers necessary curricular revisions, temporal addresses the accelerated rate of change, ethical dimensions address issues of academic integrity and bias, and the affective dimension focuses on learner and educator emotional responses and engagement with the new technologies and processes.

Successful integration of artificial intelligence within educational environments necessitates a shift from the traditional student role of passive recipient to active participant and co-creator in the change process. This requires fostering a partnership between students and educators, where students contribute to the design, implementation, and evaluation of AI-driven tools and pedagogical approaches. Specifically, students should be involved in identifying learning needs that AI can address, providing feedback on AI tool usability and effectiveness, and collaboratively developing new learning activities that leverage AI capabilities. This co-creation model ensures that AI integration is aligned with student needs and promotes a sense of ownership and agency, ultimately enhancing learning outcomes and fostering innovation.

The Cultivation of Provisional Knowledge: Collaborative and Humble Inquiry

Humble Inquiry represents a powerful shift in how institutions navigate change, advocating for an approach rooted in intellectual humility rather than assumed expertise. This methodology prioritizes understanding local contexts and actively seeking input from those directly involved, fostering collective knowledge-building as a core principle. Rather than imposing top-down solutions, Humble Inquiry champions a process of shared exploration, recognizing that valuable insights often reside within the experiences and perspectives of individuals closest to the challenges. By embracing this mindset, organizations can move beyond simply implementing change to genuinely understanding and adapting to evolving needs, ultimately creating more effective and sustainable outcomes through a collaborative spirit of inquiry.

Collective Inquiry provides a deliberate framework for tackling complex challenges and fostering innovation through a shared pursuit of understanding. Rather than relying on top-down directives or isolated expertise, this approach centers on collaborative exploration, where diverse perspectives are actively sought and integrated. Participants engage in a cyclical process of questioning, observing, reflecting, and revising, building knowledge collectively and iteratively. This methodology emphasizes that solutions often emerge not from individual brilliance, but from the synthesis of multiple viewpoints and the rigorous testing of assumptions within a supportive environment. By prioritizing shared learning and distributed cognition, Collective Inquiry empowers groups to navigate uncertainty and generate more robust and creative outcomes than traditional problem-solving methods.

Institutions seeking to meaningfully integrate artificial intelligence into higher education must prioritize the development of faculty as active agents of change. Rather than simply implementing AI tools from the top down, a supportive environment that cultivates humble and collective inquiry empowers educators to investigate how these technologies can best serve their specific pedagogical goals and student needs. This approach moves beyond technical training, fostering a culture where experimentation, shared learning, and open dialogue guide the integration process. Consequently, AI becomes a tool for genuine enhancement, adapting to the unique contexts of teaching and learning rather than dictating a one-size-fits-all solution, and ultimately ensuring that innovation is driven by those closest to the students.

The pursuit of institutional stability, so often lauded, feels increasingly like a postponement of the inevitable. This paper rightly frames generative AI not as a problem to solve but as an ‘arrival technology’ – a force demanding adaptation, not adoption. It recalls Schrödinger’s observation, “As far as I can see from my experience, people are quite willing to accept the consequences of their acts, but they are not willing to accept the responsibility for them.” Institutions, too, often react to the effects of technological shifts without reckoning with the fundamental need to evolve their core structures. Long stability, therefore, isn’t a sign of success, but potentially a hidden disaster – a system that has become brittle and unresponsive, poised to fracture under the weight of unforeseen circumstances. The framework proposed here isn’t about building change, but cultivating the conditions for it to emerge – recognizing that systems don’t fail, they evolve into unexpected shapes.

The Shape of What Remains

This exploration of institutional adaptation in the face of generative AI reveals less a solution than a careful charting of the inevitable currents. The framework proposed isn’t a scaffold to build resilience, but a weather vane to measure the coming storms. Each elegantly-defined stage of adaptation implicitly acknowledges the limits of foresight; every attempt to anticipate pedagogical shifts risks enshrining obsolescence. The real work lies not in predicting the future of STEM education, but in cultivating a tolerance for constant, unpredicted re-calibration.

The emphasis on ‘arrival technology’ is crucial, though it merely names the discomfort. The belief that institutions can ‘adopt’ such forces-integrate them neatly into existing structures-is a phantom hope. What will persist is not the framework itself, but the questions it raises. Specifically: how does an institution prepare not for a single disruption, but for a sustained state of becoming-obsolete? How does faculty development shift from skill-acquisition to the cultivation of epistemic humility?

The field now faces a choice. It can pursue ever-finer granularity in models of change, chasing a perfect map of an inherently chaotic system. Or it can embrace the inherent uncertainty, focusing instead on building systems that fail gracefully-that are capable of absorbing disruption, of learning from their own inevitable shortcomings. The latter path demands less architectural ambition, and more attentive listening to the whispers of entropy.

Original article: https://arxiv.org/pdf/2605.12757.pdf

Contact the author: https://www.linkedin.com/in/avetisyan/

The Inevitable Erosion of Institutional Memory

Building on Shifting Sands: Established Foundations for Change

Adapting IE Protocols for an Accelerating Entropy

The Cultivation of Provisional Knowledge: Collaborative and Humble Inquiry

The Shape of What Remains

See also: