While the promise of artificial intelligence often centers on labor-saving automation, a new study from researchers at Berkeley Haas suggests a more taxing reality. Analyzing employees at a U.S. tech firm, the research highlights how Large Language Models (LLMs) often act as "partners" that drive workers to manage multiple active threads simultaneously.

This shift introduces a frantic rhythm where developers might manually write code while an AI generates alternatives in the background, or revive deferred projects simply because the tool makes them feel achievable. This "multi-threading" leads to frequent attention switching—a known drain on mental energy—and a growing backlog of open tasks that keep the brain in a state of high alert.

The ease of generating results with "just one more prompt" can become addictive, leading to work sessions that deplete mental reserves within hours. This phenomenon suggests we are disrupting decades of established intuition regarding sustainable work-life balance and the limits of cognitive load.

To counter these effects, organizations are encouraged to build a formal Agent Practice. This framework, which structures how tools are deployed, aims to distinguish between genuine, sustainable productivity gains and the unsustainable intensity of constant cognitive juggling, ensuring that technological integration does not come at the cost of human well-being.

While the promise of artificial intelligence often centers on labor-saving automation, a new study from researchers at Berkeley Haas suggests a more taxing reality. Analyzing employees at a U.S. tech firm, the research highlights how Large Language Models (LLMs) often act as "partners" that drive workers to manage multiple active threads simultaneously.

This shift introduces a frantic rhythm where developers might manually write code while an AI generates alternatives in the background, or revive deferred projects simply because the tool makes them feel achievable. This "multi-threading" leads to frequent attention switching—a known drain on mental energy—and a growing backlog of open tasks that keep the brain in a state of high alert.

The ease of generating results with "just one more prompt" can become addictive, leading to work sessions that deplete mental reserves within hours. This phenomenon suggests we are disrupting decades of established intuition regarding sustainable work-life balance and the limits of cognitive load.

To counter these effects, organizations are encouraged to build a formal Agent Practice. This framework, which structures how tools are deployed, aims to distinguish between genuine, sustainable productivity gains and the unsustainable intensity of constant cognitive juggling, ensuring that technological integration does not come at the cost of human well-being.