Artificial intelligence is rapidly becoming embedded in how we learn, work, communicate, decide, and govern. Much of the contemporary discussion surrounding AI focuses on fairness, transparency, accountability, explainability, safety, and regulation. These are important concerns and have significantly advanced the field of AI ethics and governance.
However, while following these developments, I found myself increasingly drawn to a different question:
What happens if AI systems become remarkably successful?
Not what happens if they fail, but what happens if they become so effective, accessible, and integrated that they mediate an increasing proportion of human cognitive and societal activities.
This question became the starting point for my article,
Ethical and Sustainable AI Mediation: A Generational Socio-Technical Framework of Silent Cognitive Erosion and Cognitive Sustainability, recently published in AI and Ethics.
https://link.springer.com/article/10.1007/s43681-026-01202-3
The idea emerged gradually through observations from education, governance, organizational systems, and the evolving information ecosystem. As generative AI tools became capable of assisting with information retrieval, writing, summarization, analysis, interpretation, and decision support, I began to wonder whether discussions about AI governance were focusing predominantly on governing intelligent systems while paying comparatively less attention to preserving the human capacities necessary to govern those systems.
Historically, societies have developed institutions, educational systems, and governance structures that depend upon human judgment, critical evaluation, contextual understanding, and oversight. Yet as AI-mediated assistance becomes increasingly integrated into everyday activities, it raises important questions about the long-term sustainability of these capacities.
This line of thinking led me to develop the concept of
Silent Cognitive Erosion (SCE).
The term does not imply a dramatic technological failure or a sudden decline in human intelligence. Rather, it describes a gradual and often unnoticed socio-technical process through which sustained AI mediation may weaken the cognitive, informational, educational, institutional, and governance capacities necessary for independent judgment, critical evaluation, interpretive autonomy, and meaningful human oversight.
The word “silent” is particularly important. Many technological risks are visible and immediate. Silent Cognitive Erosion concerns changes that may emerge slowly beneath conditions of apparent success. AI systems may continue improving efficiency, convenience, accessibility, and productivity while deeper capacities associated with verification, reflection, deliberation, and oversight become increasingly dependent upon technological mediation.
At the same time, I did not want the framework to become an argument against AI. AI possesses enormous potential to enhance learning, accessibility, innovation, healthcare, organizational effectiveness, and public services. Therefore, focusing exclusively on erosion would provide only part of the picture.
This realization led to the development of a complementary concept:
Cognitive Sustainability (CS).
Cognitive Sustainability represents the capacity of individuals, institutions, and societies to preserve and regenerate the cognitive, informational, educational, and governance conditions necessary for meaningful human participation in increasingly AI-mediated environments.
In many ways, Cognitive Sustainability became the normative foundation of the framework. It shifts the discussion from fear of technological change toward responsible and sustainable co-evolution between human societies and intelligent systems.
Another important aspect of the research was the adoption of a generational perspective. Many discussions about AI focus on immediate outcomes, short-term risks, or near-term regulatory responses. Yet some of the most important societal consequences of technological transformation unfold gradually across decades and generations.
Educational practices, institutional cultures, governance capacities, and informational ecosystems evolve slowly. Decisions made today may influence the capabilities available to future generations. This realization motivated me to examine AI mediation not merely as a contemporary technological phenomenon but as a long-term socio-technical sustainability challenge.
The framework therefore introduces a generational lens for exploring how differing balances between erosion-oriented and restorative dynamics may influence the future trajectory of AI-mediated societies.
Developing this work required integrating insights from multiple disciplines, including AI ethics, governance studies, socio-technical systems theory, distributed cognition, resilience research, information studies, and educational theory. One of the most rewarding aspects of the project was discovering how these diverse perspectives could be connected within a common framework.
Ultimately, the article argues that the future of AI governance depends not only on the capabilities of intelligent systems but also on the capacity of societies to sustain the human and institutional foundations necessary for meaningful oversight, adaptive governance, and democratic participation.
As AI continues to evolve, questions of sustainability may become as important as questions of capability.
My hope is that the concepts of Silent Cognitive Erosion and Cognitive Sustainability stimulate further discussion, empirical research, and interdisciplinary collaboration on what it means to create not only more intelligent technologies, but also more resilient and sustainable human futures.