New research on cognitive offloading and a U.S. survey of AI checking habits raise hard questions for leadership development
Managers and executives rely on artificial intelligence (AI) to check their writing more often than the employees they supervise, according to a survey, as separate research links heavy AI use to weaker independent thinking.
The 2026 survey by Omni Calculator found 54% of managers and 48% of executives run 70% or more of what they write past AI before sending it, compared with 30% of experienced individual contributors.
The findings follow an analysis by The New York Times of three recent studies showing people who hand the thinking process to AI perform well in the moment but struggle once the tool is removed.
The managerial paradox in AI checking
Omni Calculator surveyed 705 employed American adults who use AI for writing. It found 81% of executives and 78% of managers say AI has changed whether or how they send a message.
Both groups are roughly twice as likely as entry-level individual contributors, at 13%, to send a document specifically because AI confirmed it was ready, according to the report.
The report's authors call the pattern a managerial paradox, noting AI checking is not limited to junior staff. They suggest it may reflect the volume of high-stakes communication leaders handle, such as performance reviews and messages to direct reports.
What the learning research shows
The first study in the Times analysis – by researchers from the University of California, Irvine (UC Irvine) and education company McGraw Hill – examined a decade of data from McGraw Hill's online math platform, starting in 2015. The preprint covers 3.2 million learning interactions.
After ChatGPT launched in 2022, students spent sharply less time on text-based problems that could be pasted into a chatbot, and their scores on those problems rose. On proctored tests away from their computers, results fell below pre-AI levels, the Times reports.
A second study, prepared for the upcoming Conference on Language Modeling, found adults who grew used to automating logical and analytical tasks were more likely to abandon hard questions they could otherwise solve.
"Part of how AI is making us dumber," says Adam Green, a cognitive neuroscientist at Georgetown University in Washington, D.C., is "undercutting the development of learning how to think in younger people that now have an alternative."
Using AI as a tutor, not a shortcut
The third study – a preprint released in July 2026 by professors at Middlebury College in Vermont – had more than 200 undergraduates write an essay and take a proctored multiple-choice test, with one group allowed to use AI.
A week later – with everyone working unassisted – students who had used AI to draft entire essays saw their test scores drop. Those who had used it like a tutor to explain complex concepts or clarify ideas kept high scores and wrote stronger essays, the Times reports.
None of the three studies examined working managers. The research adds to recent reporting that hiring managers prioritize judgment-based oversight of AI outputs over technical skill, and to questions about whether middle managers are being left behind in AI transformation.
What employers can do about managers' AI use
Clear rules on how leaders use AI are a starting point, alongside efforts to redesign the work itself so AI supports productivity. A KPMG in Canada survey of 2,239 employees, conducted in August 2025, found just 29% say their employer has a comprehensive AI policy, while 40% don't know what AI controls are in place at their organization.
Training should focus on scrutinizing AI output, not just producing it. A 2025 study by Microsoft Research and Carnegie Mellon University of 319 knowledge workers found higher confidence in generative AI was linked to less critical thinking, while higher self-confidence was linked to more.
Assessing leaders' judgment without AI is a third option. Research and advisory firm Gartner predicts that through 2026, the atrophy of critical-thinking skills from generative AI use will push 50% of global organizations to require "AI-free" skills assessments. The firm expects hiring to place more weight on candidates' ability to solve problems, weigh evidence and make judgment calls unaided.
