People are using artificial intelligence. A lot of them are keeping it quiet.
That’s the surprising finding laced throughout a new study of 5,000 American desk workers who use AI at work. It comes to us from Charter, a research and media organization focused on the future of work.
Overall, 35.9 percent of workers admitted they’ve concealed their use of artificial intelligence from a manager, colleague, or client.
Drilling down:
Among managers, 43.7 percent say they have concealed AI use.
For executives, the number is 37.1 percent.
And pulling up the rear, 27.1 percent of individual contributors say they’ve concealed their AI use.
Also — and maybe this should have been the lead — 32.8 percent of executives said they trusted their own judgment less than they trusted AI. And 34.8 percent of managers said the same.
That’s about twice the rate for individual contributors or line workers, among whom 16.3 percent said they trusted AI more.
As Emily Peck put it in reporting on this study at Axios:
“It’s a worrying development. Leaders are hired for their experience and knowledge. Outsourcing that to AI raises the risk of making decisions that don’t line up with reality, which then reverberate across companies and beyond.”
Now, perhaps you can understand why workers give in to the temptation to use AI and not tell anyone.
Among workers who had concealed their AI use:
49.3 percent said they worried about appearing less capable or less competent.
35.5 percent worried that they would look lazy.
26 percent of workers said they simply didn’t know if their organizations allowed AI use.
Some of the anonymous study participants’ comments shed light on other concerns:
An engineer in the IT industry: “I am worried about it completely replacing me … in my job, my particular job of coding, it could easily. Upper management could feel it could actually replace me, and I would not be needed at all.”
A manager in the healthcare industry: “You have to do more with less, and you have to challenge [your team] to do the same thing. You can see the toll it takes on other people.”
A financial analyst in the hospitality industry: “I use it out in the open. We just don’t talk about it.”
A customer service representative in the insurance industry: “I don’t really discuss what I do … because I just don’t see how it would be supported.”
All of which leads to one other takeaway: the toll that angst over AI, coupled with using it but carrying it like some kind of open secret, can take on employees.
This part gets a bit statistical:
The report says 39.3 percent of workers report that AI has reduced their burnout. Meanwhile, 13.8 percent say AI has increased their burnout. (The rest reported no meaningful change.)
But those who reported increased burnout were 28.5 percentage points more likely to have concealed their AI use than those reporting decreased burnout.
Basically, more workers said AI reduced their burnout than increased it.
But increased burnout went hand in hand with hiding AI use.
Maybe it’s time to come clean?
Thank you for reading Understandably! If someone shared this with you, please sign up here for a free or paid subscription. I’m also grateful for one-off support (via Buy Me a Coffee) that helps me keep this whole thing going!
Other things worth knowing …
ABC News: Alexander Darchiev, Russia’s ambassador to the United States, died Wednesday at the age of 66, according to the Russian Foreign Ministry. The diplomat had headed the Russian Embassy in Washington since March 2025. No cause of death was immediately announced. The Foreign Ministry announcement said Darchiev “passed away suddenly.”
WSJ | CNBC: Nvidia’s board was reportedly preparing to oust CEO Jensen Huang, its founder and chief executive after he had two extramarital affairs, including one with a suspected Chinese spy, the Wall Street Journal reported. The swirl of sex-and-spying allegations is raising a worrying question: Could the company’s trouble spread through the tightly connected financial network it has built to support the artificial intelligence boom? Nvidia has made hundreds of billions of dollars in financial commitments to other firms to help the industry grow, earning it the nickname the “central bank of AI.”
AP: Democrat Scott Colom has defeated Republican Cindy Hyde-Smith after a prolonged ballot counting process, giving Mississippi its first Black U.S. senator since Reconstruction. Colom’s narrow victory is a seismic shift in Mississippi, which has been controlled mostly by white conservatives for generations. Irony: President Joe Biden nominated Colom to be a U.S. District Court judge in Mississippi in 2022; Hyde-Smith blocked his confirmation.
Axios: Stephen Miller hopes to use his new role as White House chief of staff to win back young and minority men who helped elect President Trump in 2024 but swung toward Democrats in the midterms, two people familiar with the administration’s plans tell Axios. Young, working-class and minority men fled the GOP in droves in last week’s elections, amid high prices fueled by Trump’s tariffs, the war in Iran, and concerns about AI’s impact on the job market.
WSJ: Cranbury, N.J., established in 1697, drips with charm. White clapboard houses on Main Street proclaim their vintage—”Built by John Nesbit 1804” reads one plaque—and locals proudly share that George Washington and Alexander Hamilton met here in 1778 ahead of the Battle of Monmouth. But now this town of about 4,000 people, is reeling from a distinctly modern catastrophe: a cyber theft to the tune of $26 million after they were tricked into wiring money to the wrong place for a land purchase.
NYT: Defense Secretary Pete Hegseth was scolded on Wednesday at the Tomb of the Unknowns by an Army sentinel after laughter broke out among a group that accompanied him. Mr. Hegseth, his wife and some advisers had been posing for pictures and chatting in front of the tomb, also called the Tomb of the Unknown Soldier, a solemn memorial at Arlington National Cemetery for military members who were killed in battle and remain unidentified. “All visitors will maintain an atmosphere of silence and respect at all times,” the sentinel shouted toward Mr. Hegseth and others as they were leaving.
PBS: The historic ocean liner SS United States was to be intentionally sunk Wednesday off the coast of the Florida panhandle. In its heyday, the ship crushed the trans-Atlantic speed record, which it still holds. Carrying presidents and Hollywood stars, the nearly 1,000-foot vessel was a technical marvel of its time. Resting at 180 feet below the surface of the water, the SS United States will join more than 500 artificial reefs in Okaloosa County, which the county relies on to support the underwater ecosystem.
Thanks for reading. Photo by Alonso Reyes on Unsplash. I wrote about some of this at Inc.com. See you in the comments!
I think AI is great for some things, but should be used sparingly for others. I just had this conversation with my director yesterday.
I have a client with multiple divisions who use a software system we provide, to manage their storage with us. They are disposing of hundreds of items to make room for some new things, but the person in charge wants each division to pay for their own disposal. It's not an easy task to figure this all out. 177 pages of items and categories, some stay, some are red lined to go. I put it in AI and gave it some prompts. The result was a spreadsheet showing who owned what assets in order of quantity. Then it gave another spreadsheet on only what was being discarded and who owned what percentage of that. I sent it to the client so that she can consolidate some of the divisions and get back to me with how she wants us to invoice them. It took 4 minutes.
I'd be working on it into 2027 if I had to cull through it all by hand.
On the flip side, I'm not a fan of people going in and saying "write an email to a prospect with the following parameters" and AI spits it out with emdashes, strange words and marks that a human would never use. I do write emails and tell AI to make it more geared to a type A personality, or more conversational, less wordy, etc. and then I edit again.
Also, AI tends to write in the past; "I wanted to reach out to you". That's not how I write, and it messes with my OCD.