{
  "id": 20334599,
  "title": "Catching AI's mistakes is a mental burden—and a drag on productivity",
  "url": "https://urgent.news/2026/11/16/catching-ais-mistakes-is-a-mental-burden-and-a-drag-on-productivity-20334599",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-16T21:00:07.000Z",
  "source": {
    "name": "Phys.org",
    "slug": "phys-org",
    "url": "https://phys.org/news/2026-11-ai-mental-burden-productivity.html"
  },
  "original_language": "en",
  "account": "AI-generated content can boost productivity, but the gains are often offset by the time and mental effort required to verify its accuracy and appropriateness. This verification burden, a cognitive cost, arises from the responsibility-authority double bind - automation grants practical authority, yet responsibility remains with the human. Cognitive engineers study how people collaborate with automated decision-support systems, and generative AI verification is a modern manifestation of this issue. Verification is a distinct task from creation, with research showing that people recall their own work better than information they merely read. Easy-to-read AI responses are perceived as more credible, leading people to consider them more likely to be true. However, humans often fail to thoroughly vet AI responses, instead weighing their own costs and benefits to decide when to check. Overreliance on AI output is challenging to resist, and even adding explanations of an AI's reasoning does not eliminate the quandary people face in committing time and energy to verification. Some studies found productivity gains from using generative AI, but a randomized trial with experienced software developers showed no improvements, and completion times even increased by 19% when a verification step was included. In consulting teams, using AI outside a person's area of expertise resulted in more people reaching the wrong conclusions. Verification can feel burdensome in terms of cognitive load, time, and motivation because it's fundamentally different from creation. Workers must decide what to check and how, and they're uncertain about potential errors, where they might be, and their forms. This resembles airport security screening, where rare threats are often missed because inspectors assume they're not present. Verification is more burdensome when AI generates work beyond the worker's ability to evaluate. A study showed AI reasoning helped with easy mistakes but not hard ones, when help was most needed. More work is required to fix identified mistakes, leading people to be less likely to fix them. Verification also carries a motivational burden; creation provides visible progress, while verification may find no errors, making verification feel unproductive. Over time, workers may become less vigilant when automation usually performs well. These burdens result in a skill swap, with critical thinking shifting from production to verification. Workers need to learn how to recognize AI agents' capabilities and limits, including omissions and biases. Companies must provide time, training, and authority for workers to verify AI-assisted output to ensure trustworthy work at a lower total cost. Measuring productivity gains should consider the entire human-AI workflow, not just how quickly AI produces a first draft.",
  "summary": "By now, almost everyone has asked an AI agent a question, only to realize that the answer seems not quite right. Something might be missing or wrong or out of date. An executive assistant told me she uses AI tools for drafting itineraries and briefings, but she's caught enough issues that she now builds time into her workflow to thoroughly check for inaccuracies and biases.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}