The time saved replacing humans with AI is reallocated to managing, training, and iterating on those agents. This is a significant, ongoing operational cost that many overlook, requiring daily attention to prevent performance degradation and ensure alignment.

Employees, burned out from the unrewarded labor of "botsitting" (managing AI), eventually hit a breaking point. This leads them to "botshit"—delivering AI-generated work they can't explain or defend. The root cause is systemic, not just individual laziness.

Andrew Wilkinson reveals the hidden cost of using AI agents for automation. He spends the majority of his time debugging and improving them, with only a small fraction dedicated to actual productive output. This highlights the immaturity of current agent technology despite its power.

A Workday study reveals a critical blind spot in AI productivity metrics. While tools save time, roughly 37% of that saved time is offset by the need for rework—verifying information, correcting errors, and rewriting content. This dramatically reduces the net value and ROI of the technology.

A massive gap exists between individual productivity boosts from AI (saving 13 hours/week) and tangible organizational performance improvements. This suggests that individual gains are lost in coordination failures and hidden labor, not translating to the bottom line.

When AI empowers non-specialists to perform complex tasks (e.g., marketers writing code), it creates a new, hidden workload for experts. These specialists must then spend significant time reviewing, correcting, and guiding the AI-assisted work from their colleagues, creating a new form of operational drag.

Research highlights "work slop": AI output that appears polished but lacks human context. This forces coworkers to spend significant time fixing it, effectively offloading cognitive labor and damaging perceptions of the sender's capability and trustworthiness.

While AI coding assistants appear to boost output, they introduce a "rework tax." A Stanford study found AI-generated code leads to significant downstream refactoring. A team might ship 40% more code, but if half of that increase is just fixing last week's AI-generated "slop," the real productivity gain is much lower than headlines suggest.

Analysis of AI-related work reveals an "exhaustion multiplier." The most draining activity for employees is not debugging or re-prompting, but the repetitive task of providing AI with basic context, like authoritative documents. This is seen as a fundamental failure of the tool itself.