Salesforce CEO Marc Benioff said in a recent interview that AI now does up to 50% of all work at the company, in key functions like engineering, coding, and customer support. In May, Microsoft CEO Satya Nadella said 20% to 30% of the tech giant’s code is now written by AI coding assistants. And in April, Google CEO Sundar Pichai said over 30% of code at Google is now generated by AI.
It’s the latest CEO flex: Citing numbers showing that AI is doing heavy lifting internally. The move presents the company as being ahead of the AI curve—and invariably grabs the attention of people who matter.
Investors hear the magic words that the business is on track to save money, presumably accomplished, but rarely explicitly stated, through future job cuts. It also signals to the clients of the Big Tech companies making the pronouncements that they should open their wallets, pronto, to incorporate more AI into their operations, or risk falling behind.
But how significant these CEO flexes from Salesforce, Google, and Microsoft ultimately are is difficult to know. The metrics cited seem precise, yet when asked, their spokespeople declined to provide any details about how the numbers were calculated or how they defined the work that they claim AI has done.
“The truth is, we don’t yet have a common framework for measuring what ‘percent of work’ really means in the age of AI,” said Malvika Jethmalani, founder of human capital advisory firm Atvis Group, in a message to Fortune. “Are we counting lines of code, tasks completed, hours saved, or business outcomes influenced?”
For example, on the Lex Fridman Podcast, Pichai explained that AI coding tools like Goose increase the productivity of Google’s engineers by approximately 10%, calculated by tracking hours saved weekly. However, that number assumes that engineers use those extra hours to work more rather than slack off.
The metric that AI tools are responsible for generating 30% of all new software code at Google is equally fuzzy. Does the number refer to raw lines of code that programmers suggest, committed code, or code accepted into production?
Benioff was even more vague. In the interview, he described AI’s ability to do up to half the work at his company as a “digital labor revolution,” but he didn’t clarify what “work” means in this context. For example, he mentioned using AI to co-author Salesforce’s corporate plan, but did not detail what that plan was or how much AI contributed. Did it suggest the outline for the plan or did it contribute parts of the text? Were its suggestions retained in the final document?
Other experts, however, say comments like those by Benioff are not a flex at all, but simply the reality of AI changing the world of work. Holger Mueller, vice president and principal analyst at Constellation Research, told Fortune that generative AI will massively change the work of the knowledge worker—though, in his view, without generating mass layoffs. “With developed countries facing a labor and talent shortage, more automation is the biggest promise to deliver long-term competitiveness,” he said.
But while there may be some truth in CEO statements about how much work is already being done by AI, the numbers are very vague and abstract, said Netherlands-based occupational psychologist Marais Bester. “We often see that CEOs use this type of language,” he said. “I think it’s also sort of an indicator to employees, saying, you better watch your back, you better perform.” From a business psychology standpoint, that’s not good leadership, he added.
“I was actually a bit disappointed by that comment,” he said, referring to Benioff’s statement,“because I don’t think that we’ll ever move towards a space where it will only be AI technologies being utilized as employees within an organization. There will be complementary relationships between human employees and technology.”
The flex can even cause anxiety among employees who hear it as “we’re automating you out,” Jethmalani said. “That kind of message can erode trust and undermine adoption at a moment when we need employees to show up highly engaged and willing to experiment and innovate with AI.”
Shonna Waters, an organizational psychologist and CEO of advisory firm Fractional Insights, also pointed out that while Benioff touts how much Salesforce is using AI—and how much its clients are adopting that company’s AgentForce platform for managing AI agents—research from firms like Gartner suggests that many of these AI-driven projects are likely to fail by 2027 because their value is unclear.
“I do think that really sets the stage for companies to be really thoughtful about how they integrate AI into their organizational design,” she said, adding that companies also must deal with the disconnect between what C-suite executives say about AI and what’s actually happening on the ground.
“These leaders are making these bold claims, and employees are experiencing something pretty different,” she said. CEOs, she explained, often have more optimism about AI than employees, while employees have more angst.
The companies that will succeed, she said, will be those with “structural empathy”—that is, building systems that bring in frontline worker voices. “At the end of the day, you need the humans to still be the ones actually adopting the AI you need to bring them along with you journey and figure out how to do it in concert with them, as opposed to something you’re doing to them.”
Bester said CEOs may be using this flex as little more than a boast to competitors. They are saying “just look at us, we are ahead of the curve on this,” he said. A better message from Benioff, he said, “would have been about how by utilizing AI and with the human capital strength that we already have, we are able to do so much more than we are already doing in terms of creating efficiencies and better value for our customers.”
For now, CEOs “obviously want to show their stakeholders that they are on board with AI” and focusing on efficiency, margins, and building value for shareholders,” Bester added. “But it could potentially backfire” if organizations don’t keep in mind how they are communicating with employees.
Or perhaps, if they have to rehire humans down the line if AI proves unable to do so much work. In May, just months after touting AI’s ability to replace human workers, Klarna CEO Sebastian Siemiatkowski reversed an AI-driven hiring freeze and announced the company is adding more human staff. He told Bloomberg that Klarna is now hiring to ensure customers always have the option to speak with a real person. “From a brand perspective, a company perspective, I just think it’s so critical that you are clear to your customer that there will always be a human if you want,” he said.
