Thinking Fast, Slow—and No Longer Agency decay—the gradual erosion of independent human thought, decision-making, and action—is occurring at the scale of entire societies as artificial intelligence systems increasingly replace human judgment, according to a new analysis. The phenomenon, described as cognitive erosion operating like muscle atrophy, progresses from AI exploration to routine integration to dependency, ultimately stripping humans of the capacity to think independently. Researchers warn that this gradual disempowerment, driven by local decisions to automate for competitive advantage, is becoming an existential risk with a rapidly closing window to reverse the trend. Artificial Intelligence /us/basics/artificial-intelligence Thinking Fast, Slow—and No Longer How agency decay happens amid AI—and what we can still do about it. Posted June 4, 2026 Reviewed by Michelle Quirk /us/docs/editorial-process Key points - Agency decay is occurring at the scale of entire societies, which is gradually becoming an existential risk. - Cognitive erosion operates like muscle atrophy—invisible until you need the strength and find it gone. - We can orient ourselves toward active engagement with AI that preserves rather than depletes human capacity. Some types of loss announce themselves only in retrospect. Too late. You reach for a word in your mind, and it has gone. You try to navigate a city you once knew without your phone and realize the map has quietly dissolved from your internal landscape. You want to watch a movie and open Netflix relying on recommendations. Small disappearances, barely registered. But multiply those moments, widen the frame from one person to an entire civilization, and the urgency of this moment becomes apparent. What happens to human beings when machines do the thinking for them? We are the humans that are both subject and object in that interrogation… The Three Speeds of Human Thought Daniel Kahneman gave us the language of System 1 and System 2 https://dn790002.ca.archive.org/0/items/DanielKahnemanThinkingFastAndSlow/Daniel%20Kahneman-Thinking%2C%20Fast%20and%20Slow%20%20.pdf —thinking fast and thinking slow. System 1 is automatic, intuitive, and effortless. System 2 is deliberate, analytical, and costly. Both matter. Both are trained by experience and sharpened by use. Increasingly, studies now refer to a third mode, artificial https://www.researchgate.net/publication/399711077 Thinking-Fast Slow and Artificial How AI is Reshaping Human Reasoning and the Rise of Cognitive Surrender . Sadly, that might not be the end of our cognitive decline https://www.psychologytoday.com/us/basics/mild-cognitive-impairment . Beyond fast, slow, and artificial lurks the dark territory of not at all. Delegating the cognitive act itself to a machine that is faster, tireless, and statistically fluent—and absorbing its output as if it were our own. This is a description of a trajectory. Curious artificial intelligence https://www.psychologytoday.com/us/basics/artificial-intelligence AI exploration morphs into routine integration, routine integration morphs into dependency, and dependency—unchecked—gradually shades into the gradual erosion of the capacity to think, decide, and act independently. Agency decay https://www.psychologytoday.com/us/blog/harnessing-hybrid-intelligence/202506/the-risk-of-agency-decay-amid-ai-use , mission completed. From Your Mind to Our World Agency decay is worrisome enough as a personal story. But the same erosion is occurring at the scale of entire societies https://arxiv.org/abs/2501.16946 , which is gradually becoming an existential risk. We are navigating a Hybrid Tipping Zone https://www.psychologytoday.com/us/blog/harnessing-hybrid-intelligence/202509/the-hybrid-tipping-zone-0 . And the window of opportunity to shift the trend is closing, fast. Gradual disempowerment refers to the incremental loss of human influence over the large-scale systems—economies, governments, cultural production—on which civilizations depend. This is distinct from the dramatic AI-takeover scenarios of science fiction. No robot uprising. No sudden betrayal. Instead, something far harder to pinpoint and resist: Each small replacement of human judgment by machine judgment makes sense locally, in isolation, under competitive pressure. Companies automate because the alternative is falling behind. States adopt algorithmic governance because it is cheaper. Cultural platforms optimize for engagement because attention https://www.psychologytoday.com/us/basics/attention is the currency. Nobody chooses disempowerment. It accumulates. Societal systems have historically remained aligned with human interests primarily because they needed humans to function. Economies needed workers. Governments needed voters. Cultures needed human creators. Strip away that structural dependency, and the alignment becomes precarious. A state funded by taxes on AI profits has considerably less incentive to represent its citizens than one funded by their labor. What makes this pattern pervasive is that its drivers reinforce each other. And none of them require malice. This dynamic requires only incentive structures pointed in one direction. The Speed Trap Inside the Mind System 1 thinking—the fast, intuitive kind—is built from accumulated experience. Years of pattern recognition https://www.psychologytoday.com/us/basics/apophenia , emotional calibration, embodied skill. System 2 thinking—slow, deliberate, analytical—requires effort and the willingness to be wrong. Both depend on a history of actually doing the cognitive work. What agency decay does, essentially, is interrupt that history. You never quite accumulate the experience that fast thinking would require, because the machine handled the repetitions. You never develop the tolerance for effortful struggle because the machine handled the difficulty. The result is a reduction in cognitive sovereignty https://www.psychologytoday.com/us/blog/harnessing-hybrid-intelligence/202602/are-we-cruising-toward-cognitive-capitulation —the capacity to think through problems with your own resources, to trust your own judgment, to function when the machine is absent, or wrong. Cognitive erosion https://www.cigionline.org/articles/the-silent-erosion-how-ais-helping-hand-weakens-our-mental-grip/ operates exactly like muscle atrophy—invisible until the moment you need the strength and find it gone. The brain is a muscle, use or lose it. We Must Move Beyond the Aspiration for Autonomy to Action for Its Sake Noticing that we are at risk and resisting that danger are themselves cognitive and political acts—and both become harder as our algorithmic disempowerment accumulates. A population that has individually and collectively outsourced its information synthesis, its creative expression, and its deliberative reasoning to AI systems is not equipped to recognize what has happened until it is too late. It is also less capable of organizing a response. Agency erosion is self-obscuring. Intelligence https://www.psychologytoday.com/us/basics/intelligence Essential Reads Add that feedback loops operate across domains simultaneously—economic, cultural, political. At the time of writing, no one has a concrete, plausible plan for stopping human disempowerment amid AI. What exists are frameworks, warnings, and the faint beginning of governance discussions. The A-Frame: A Practical Takeaway Existential risk framing can feel paralyzing. So here is something practical—a way to orient yourself, your team, your institution toward active engagement with AI that preserves rather than depletes human capacity. The A-Frame offers four anchors. Awareness: Name the pattern before it names you. Agency decay does not announce itself. It shows up as the slight discomfort of writing without an AI completing your sentences. As the mild anxiety https://www.psychologytoday.com/us/basics/anxiety of navigating a problem without a prompt. Pay attention to those moments. They are diagnostic signals. The first step is simply to see the erosion happening, in yourself and in the systems around you. Appreciation: Recover a genuine sense of what human cognition https://www.psychologytoday.com/us/basics/cognition is and does. Slow thinking is the substrate of judgment. Memory https://www.psychologytoday.com/us/basics/memory is the foundation of meaning-making and identity https://www.psychologytoday.com/us/basics/identity . Creative struggle is where original thought is actually forged. Appreciation means valuing the process, not merely the output. Acceptance: AI is here, accelerating, and neither panic nor wholesale surrender is a useful orientation. Acceptance means inhabiting the tension—using AI tools with clear-eyed awareness of what they cost, rather than either refusing them or disappearing into them. Sequence matters. Structure the process so the human mind does the foundational work, and the machine supplements what follows. Accountability: This is where individual practice becomes structural demand. Ask your employer what happens to the cognitive skills of workers whose roles are being automated. Ask your children’s school whether the curriculum is being designed to maintain independent thought. Ask your government what frameworks exist for measuring human capability alongside AI productivity https://www.psychologytoday.com/us/basics/productivity . Accountability means refusing to let the gradual become invisible. Thinking fast and slow or no longer. We still have a choice to escape the draw of quiet convenience. Let's make it before we no longer see the need.