· Desk editor: Sara Farooq · · Single source ·Updated

Workers face mental burden verifying AI output accuracy

Cognitive engineers report that employees now spend significant time checking generative AI responses for errors and biases in modern office workflows.

Catching AI's mistakes is a mental burden—and a drag on productivity
File photo Catching AI's mistakes is a mental burden—and a drag on productivity Photo: Phys.org

Verification adds cognitive load to work

An executive assistant uses AI tools for drafting itineraries but builds extra time into her workflow to check for inaccuracies. This verification burden counters productivity gains because workers must determine if output is accurate and appropriate. Cognitive engineer David Woods described this problem in 1985 as the responsibility-authority double bind where automation holds authority yet humans keep responsibility.

Research shows completion times increase

A randomized trial of experienced software developers found no productivity gains when a verification step was included. Completion times increased by 19 percent because developers believed they were faster while using the tool. An extensive examination of consulting groups revealed that applying AI beyond internal knowledge led to a greater number of individuals arriving at incorrect conclusions.

Simple explanations fail to help

One research team found adding simple explanations of reasoning did not eliminate the quandary regarding time and energy needed to check. Observers often categorize agents as able or unable instead of seeing them with subtlety. A meta-analysis examining 86 studies found that confirming fresh data differs from generating it.

Reported by one outlet

Only one outlet has published this. Nothing here has been checked against a second report, so read it as that outlet's account and follow the link below for the original.

Reported by

1 independent outlet. Headline as published. Links open the original report.