Cognitive debt is a concept popularized by Simon Willison in February 2026 describing the hidden cost of AI-accelerated software development. Unlike traditional technical debt, which accumulates in the codebase itself, cognitive debt accumulates in the minds of developers – they lose the mental model of what their software does and how it works.
Willison defines cognitive debt as the consequence of AI-accelerated development outpacing human comprehension: “We take on cognitive debt when AI accelerated development means we can no longer maintain a robust mental model of what our software does and how it works.”1)
The term gained traction through Margaret-Anne Storey's explanation, which Willison called “the best explanation of the term cognitive debt I have seen so far”: “Cognitive debt, a term gaining traction recently, instead communicates the notion that the debt compounded from going fast lives in the brains of the developers and affects their lived experiences and abilities to go fast or to make changes.”2)
Technical debt lives in the code – messy architecture, shortcuts, duplicated logic. Cognitive debt lives in developers' heads. Even if AI produces code that is clean and well-structured, the humans involved may have “simply lost the plot and may not understand what the program is supposed to do, how their intentions were implemented, or how to possibly change it.”3)
The distinction matters because cognitive debt can paralyze a team even when the codebase looks reasonable.
Willison identifies the core dynamic: “Agents let us move so much faster, but this speed also means that changes which we would normally have considered over the course of weeks are landing in a matter of hours. It is so easy to let the codebase evolve outside of our abilities to reason clearly about it. Cognitive debt is real.”4)
Mario Zechner, creator of the Pi agent framework used by OpenClaw, describes the dynamic bluntly: “You have zero fucking idea what is going on because you delegated all your agency to your agents. You let them run free, and they are merchants of complexity.”5)
Storey describes a student team she coached that hit a wall by weeks 7-8 of AI-accelerated work. They could no longer make simple changes without breaking something unexpected. The team initially blamed technical debt, but the real problem was deeper: “No one on the team could explain why certain design decisions had been made or how different parts of the system were supposed to work together. They had accumulated cognitive debt faster than technical debt, and it paralyzed them.”6)
Willison acknowledges experiencing it firsthand: “I have experienced this myself on some of my more ambitious vibe-code-adjacent projects. I have been experimenting with prompting entire new features into existence without reviewing their implementations and, while it works surprisingly well, I have found myself getting lost in my own projects.”7)
Not all AI-generated code creates cognitive debt. Simple code – fetching database data and outputting it as JSON – has implementations obvious enough that understanding is not critical. But when core application components become black boxes that developers do not fully understand, confident reasoning about the system becomes impossible, making planning new features harder and eventually slowing progress8).