The Silent Erosion of Cognitive Agency
AI is embedding itself deeply into how we think, formulate questions, and make decisions. While touted as empowerment, this is a dangerous irony. It's leading to "agency decay." Every day, we watch drivers request directions for routes they once knew and writers open a chat window before facing a blank page. These choices appear minor, yet they fundamentally alter where knowledge resides, who exercises judgment, and what happens when the system fails. Tools may empower us to do more faster, at a lower cost. But we need to ask: What is the true cost?
Generative AI has entered the thought process itself. It is increasingly woven into the chain through which we formulate questions, interpret evidence, and decide what deserves attention. But AI-powered empowerment is an oxymoron. When individuals and institutions continuously transfer creation, judgment, execution, and information to a system, they gradually lose the ability to create, grasp, and master the processes that the tool was only meant to make better. In essence, the tool is empowered, often at the expense of those who deploy it.
The Mechanics of Agency Decay
Agency decay is not an irreversible verdict, yet. It resembles a limb kept in a cast: the capacity remains, yet strength and coordination diminish through disuse. Skills rarely disappear in one dramatic moment. Instead, they erode because the tool is always available, the shortcut is rewarded, and independent practice begins to feel inefficient.
Memory research has found a related effect. A study published in 2011 showed that when people expect information to remain available online, they remember where to retrieve it more readily than the information itself. GPS research likewise found that heavier lifetime reliance on navigation systems was associated with weaker spatial-memory performance. We did not suddenly become less intelligent; we adapted to the environment and stopped retaining what the environment promised to supply, only remembering the shortcuts to access that supply. This is the luring danger of compounding comfort. When we rely on AI to do the thinking for us, we forget how to do the thinking ourselves. If the system were to suddenly vanish, would we still have the ability to formulate those same questions and critical insights on our own? The data points to a sobering "no."
Why Automation Irony Still Holds True
The pattern we see today predates generative AI. In "Ironies of Automation," psychologist Lisanne Bainbridge showed already in 1983 that as automated systems become more reliable, human operators get less practice handling the situations for which they are still held responsible. When the system fails—and it inevitably will—control returns to the person whose readiness has been weakened by its success.
Early findings on generative AI deserve caution, yet they point in the same direction. A 2025 MIT Media Lab preprint reported weaker neural connectivity among participants writing essays with an AI assistant than among those working without one. Furthermore, Microsoft researchers surveying 319 knowledge workers found that greater confidence in generative AI was associated with less critical-thinking effort, while greater confidence in one’s own ability was associated with more. We aren't just losing practice; we are becoming less engaged with the fundamental tasks of our work.
Collective Atrophy in Our Organizations
At the organizational level, this becomes collective agency decay. Companies are not only buying software; they are embedding external AI systems into research, customer service, coding, recruitment, logistics, and strategic analysis. They transfer proprietary data and operational knowledge into tools that increasingly mediate how work is done. Over time, part of the organization’s practical intelligence migrates beyond its boundary.
The dependency is technical, commercial, and cognitive. Processes are redesigned around a provider’s models and interfaces. Staff stop practicing the underlying work. Switching suppliers then means more than just replacing a license: It requires extracting data, rebuilding workflows, retraining people, and recovering capabilities that have atrophied. The OECD work on data portability links poor interoperability to higher switching costs, while the National Institute of Standards and Technology (NIST) generative-AI risk profile explicitly advises organizations to identify overreliance on third-party data and systems, maintain fallbacks, and scrutinize vendor arrangements. If an organization cannot perform its own core functions without the AI, it has ceded more than just data—it has ceded its autonomy. Security risks can escalate when permissions are improperly inherited by such systems.
Reclaiming Hybrid Intelligence
The constructive goal is hybrid intelligence: artificial assets extending natural intelligence, rather than gradually replacing its exercise. We must consciously decide where to use AI and where to retain human sovereignty.
We cannot blindly adopt these tools simply for convenience. We must treat cognitive exertion as a form of intellectual hygiene. We should deliberately engage in the tasks that the AI handles—formulating the initial draft, structuring the argument, calculating the base probabilities—and then use the AI as a check, not as a replacement. This ensures that the capability to do deep work remains firmly within our own grasp. We should be using AI as a lever, not as a replacement for our own judgment. The long-term durability of our work relies on our ability to outthink, not just out-automate, the problems we face. The sad irony of our hybrid empowerment is only fatal if we let it be. If we recognize the decay early, we can still choose to cultivate our mastery alongside our convenience. Security incidents such as autonomous breaches demonstrate the high cost of unchecked reliance. It's time to intentionally invest in the capacity that makes us essential.