There’s a huge disconnect between the executives spamming your LinkedIn feeds with promises to go all-in on AI and the company leaders still fighting with the seventh-floor printer. Meanwhile, their employees are kinda just…doing whatever they want.

Companies like Shopify and Box have snagged headlines in the last few months for saying they are pivoting to an “AI-first” business:

• Shopify CEO Tobi Lütke told employees that everyone was expected to learn how to do their jobs with AI, and that new hires would only be approved if managers demonstrated that AI could not meet a need.
• The Norwegian hedge fund manager Nicolai Tangen, who leads the country’s massive sovereign wealth fund, told Bloomberg that he sees no future at his firm for employees unwilling to get on board.

But some of it might be about positioning (for now). Gartner Distinguished VP Analyst Arun Chandrasekaran told Tech Brew that corporate AI-first pronouncements are “a way to signal to the investors that we’re not going to be lagging behind.” Some of it is also about making it clear what kind of employees they want working for them: AI enthusiasts.

Most employees are a lot more skeptical

Research released in March from enterprise AI startup Writer found that, while 75% of C-suite execs think their company’s AI rollout in the previous year was successful, only 45% of employees believed the same.

The same survey found that 41% of younger employees have admitted to blatantly ignoring requests to adopt AI. Some workers fear they’ll eventually be replaced.

And some CEOs might be concerned if they knew how their employees are embracing AI: One report from accounting and consultant firm KPMG found that of the US employees who said they use AI in their everyday workflow, 44% reported “knowingly using it improperly,” like uploading sensitive IP to public platforms.—MM

AI in most workplaces is here to stay. But several businesses bought into the early hype that generative AI would eliminate the need for hiring people and help trim payrolls, only to learn the hard way that robots aren’t yet ready to replace humans.

AI-ers’ remorse: A recent survey from Orgvue noted that more than half of business leaders said they regretted laying off employees as a result of an AI deployment. The research also found that 40% of executives reduced staffing in order to implement AI, and 55% of those robot-lovers regretted that decision. Per S&P Global, 42% of companies abandoned their generative AI pilot projects in 2025, up from 17% last year.

To err is human, to rehire divine

Several companies quickly embraced AI as the employee of the month only to find themselves like Rose in the freezing cold Atlantic, blowing her whistle after the Titanic sank. Here are some of the companies that want people to come back:

• Klarna: CEO Sebastian Siemiatkowski said his push to use AI in a customer service role was a mistake because people would prefer to talk to humans than robots, something that eluded him even though that’s been true since the first automated voice on the other end of a phone call said, “To speak to an operator, press 0.” Siemiatkowski is not above AI replacing himself, though: You can call a hotline to give feedback to his AI-generated clone.
• IBM: After laying off 8,000 people, including many people in human resources roles who were replaced with the AskHR AI service, the company reversed course upon realizing humans may be better equipped for a job with “human” in the title.
• McDonald’s: Using AI to take orders at drive-thrus turned out to be a clown show, with the tech adding bacon to an ice cream order (among other issues).
• Duolingo: CEO Luis von Ahn made a big statement that the company was going “AI-first” and replacing contractors with AI, but he walked back that plan after facing backlash from customers.

Zoom out: The rush to get up to speed with a hot new technology was partly due to fear of falling behind competitors welcoming it with open arms, like Apple being left in the dust by its rivals on the AI front. But according to a report from The Economist, most companies aren’t clamoring for AI that’s more clever—they need tech that’s more applicable to their businesses.—DL

As AI masters more of the tasks that make up people’s nine-to-fives, some are wondering whether any career besides belly dancing has a future.

The question is most salient for young people facing an anemic white-collar job market, which has been partially attributed to AI taking over certain cubicle roles. It’s spurring some to train to be pros you call when you have a leaky toilet, as well as electricians, HVAC technicians, and elevator mechanics. The share of students studying in two-year vocational colleges rose to nearly 20% last year from 15% in 2019, per the National Student Clearinghouse Research Center.

But blue collar isn’t the only way

Predictions about knowledge workers being replaced by machines have been off the mark before. In 2016, AI research luminary Geoffrey Hinton announced that radiologists—the doctors responsible for interpreting medical imaging—would become obsolete within five years. But fast-forward to 2025:

• The scan specialists are in greater demand than ever, with the Mayo Clinic having grown radiologists’ ranks by 55% since 2016, according to the New York Times.
• While these doctors now leverage AI to help identify abnormalities in scans, they still use their human experience to interpret the results based on a deeply informed analysis of the patient’s history and advise other clinicians on care.

And tech workers might not be doomed: Despite AI taking on many entry-level software engineering tasks, many experienced developers say that they aren’t worried about automation. They believe that a deep knowledge of software and the ability to orchestrate AI agents performing complex coding tasks will remain top Indeed qualifications.

Big picture: Jobs requiring creative thinking and soft skills on top of analytical reasoning are less likely to be automated, according to career experts. And it’s become a career advice cliché, but learning to use AI makes a pro less prone to being replaced by it.—SK

As someone who recently wrapped up a job search, I know well that today’s job market is a digital jungle. You have to sign up for online job boards, tailor your profile for LinkedIn, optimize your résumé for applicant-tracking systems, and then retype all of that work experience on another page for some reason. And, before you ever sit down for an actual interview, you have to convince the algorithms you’re worthy. To do that, you need someone on the inside: Enter ChatGPT.

Fighting fire with fire

When I started searching for a new job earlier this year, I didn’t rely on AI to write my résumé or craft any cover letters (I’m a writer, after all!), but I found it was helpful to employ the tech as a second pair of eyes.

It identified typos, grammatical errors, and clunky language. Sometimes, it whiffed (its obsession with em dashes is real), but I appreciated having an editor.

ChatGPT’s biggest contribution: It spotted a font encoding problem on my résumé that undoubtedly would have kept me out of consideration for fruitful opportunities. With an increasing number of companies using AI to sort through applications, if ChatGPT flags a readability issue, it’s an issue.

Mock…Yeah!

Once I got invited to participate in the interview process, AI had something to offer there, too.

To get the job I have now, I copied-and-pasted Morning Brew’s job posting into ChatGPT and asked it to create some Brew-specific interview questions. In the past, I would’ve googled “popular interview questions” and gotten generic responses, but with ChatGPT, I had the material to stage a full-on mock interview with relevant and granular questions.

My main takeaway from using AI in a job search: Understand what AI can and can’t do, then harness it. Fight the machines with machines and hope that’s as close to the Terminator movies as all of this AI stuff gets you.—BC

Titans of industry are doing the AI equivalent of setting up their own wi-fi to avoid connecting to public networks.

To protect sensitive corporate data, business giants have largely tried to restrict their employees from using widely accessible chatbots like ChatGPT for job-related queries—especially if they’re seeking AI-generated help on confidential client work.

Instead, companies are internally developing legions of in-house bots:

In consulting, the Big Three are full-sending custom AI. At McKinsey, for example, three-quarters of the firm’s 40,000 global employees use its internal AI platform, Lilli, every month for junior-level tasks, like research and making PowerPoint slides. Kate Smaje, who leads McKinsey’s AI efforts, said this automation doesn’t “necessarily” translate to fewer human workers, but McKinsey has laid off about 5,000 employees since a few months after Lilli’s launch in 2023.

In banking, Morgan Stanley said it saved coders approximately 280,000 hours already this year after launching AI tools that can translate legacy coding into modern computer languages and standards. At Goldman Sachs, employees can generate 95% of an IPO prospectus, a document that aims to persuade new clients, in minutes using AI, CEO David Solomon said in January. That type of work used to take a team of six people more than two weeks to complete.

In retail, Walmart said it shortened its production timeline for clothes and accessories from six months to six- to eight weeks using its new AI tool called Trend-to-Product, which quickly creates mood boards from internet trends. Within the past year, Target and Amazon have both launched custom chatbots for their employees to use instead of ChatGPT.

Better make sure it’s secure: UnitedHealth Group’s Optum healthcare company got caught with its digital pants down in December, when TechCrunch reported that its internal AI chatbot, which employees asked for advice in determining claims, was publicly accessible. Optum quickly restricted the tool and said it was a demo that was “never scaled nor used in any real way.”—ML