There’s a sure wildness within the tech trade lately that each mimics earlier eras of huge modifications, like cloud computing (runaway prices within the early days), and is like nothing we’ve ever seen earlier than (document revenues accompanied by mass layoffs).
A principle doing the rounds makes an attempt to clarify the phenomenon: Tech executives, particularly CEOs, are collectively affected by delusions of grandeur because of AI. And at the very least one tech CEO has mentioned so out loud: Field founder Aaron Levie.
“CEOs are uniquely liable to AI psychosis as a result of they’re sufficiently distant from the final mile of labor that also has to occur to generate most worth with AI,” Levie wrote on X.
CEOs “play with AI,” develop a prototype, or generate a contract, to make use of Levie’s examples, after which make the leap to believing brokers can do the work.
However these top-level executives aren’t the individuals who must evaluate code, uncover bugs, and determine calls to hallucinated libraries earlier than software program is deployed. They aren’t liable for coaching AI fashions on an organization’s idiosyncratic contract phrases, nor have they got to spend days combing via contracts to seek out sneaky phrases, as Levie signifies.
In different phrases, Levie’s principle posits, CEOs don’t actually perceive processes properly sufficient to know what actually can and might’t be automated. However that lack of awareness doesn’t cease them from appearing on their beliefs.
It’s necessary to notice that Levie isn’t an AI hater. Fairly the alternative. He largely posts AI positivity on X to his 2.7 million followers, writing blogs titled, “Headless software is the future” on how software program constructed for AI brokers is the best way ahead. He additionally places his cash the place his mouth is, backing AI startups as an energetic angel investor.
So what are CEOs to do as an alternative? Levie advises CEOs to make use of AI “a ton” to actually see what it may well and might’t do, “and are available out the opposite facet with an appreciation for each the upside and the actual work.”
I’ve sufficient religion in humanity to imagine that there are CEOs on the market trying to just do that, however proper now, they appear to be within the minority.
In solely the primary 5 months of 2026, the tech trade has already had almost as many layoffs as in all of 2025: 115,430 folks have been fired from 152 tech corporations thus far in 2026, in comparison with 124,636 folks let go by 275 corporations in 2025, in accordance with trade layoff tracker Layoffs.fyi.
And the majority of corporations have pointed to AI as a cause for reducing these jobs. Many argue that the most important tech corporations are AI washingor crediting AI productiveness features prior to now or future, when different enterprise selections and metrics are actually driving the cuts.
Nonetheless, a few of these tales are shocking. Zeb Evans, the CEO of challenge administration and productiveness software program startup ClickUp, proudly declared on X that he had laid off virtually 1 / 4 of his workers — 22% — after rolling out about 3,000 AI brokers to do inside work.
Evans swore this wasn’t carried out to cut back prices. As an alternative, he desires a workforce composed of people that run AI brokers and spend their days rapidly reviewing the brokers’ work. He believes this may create a “100x org,” as he calls it.
Whereas AI is usually a very useful gizmo, the information on AI and productiveness doesn’t help such assumptions. By miles.
A meta evaluation of different analysis printed in October in UC Berkeley’s California Management Review discovered “no sturdy relationship between AI adoption and mixture productiveness achieve.”
Analysis printed in March by the National Bureau of Economic Research did conclude that AI adoption improved productiveness, however famous “a productiveness paradox, during which perceived productiveness features are bigger than measured productiveness features.”
After creating hundreds of brokers to work on duties, researchers at MIT concluded that brokers simply aren’t doing human-quality work but in lots of circumstances. They predict on the present price of LLM enchancment, fashions will “be capable of full most text-related duties with success charges of, on common, 80%–95% by 2029 at a minimally ample high quality stage.”
In different phrases, AI is on observe to carry out at base competence on most duties in about three years. These researchers imagine brokers will want one other few years to outperform people.
In the meantime, analysis printed within the Harvard Business Review confirmed that when everyone seems to be utilizing AI to provide extra stuff, the bottleneck merely shifts to executives. Their work awaits the folks that should authorize all of the stuff everyone seems to be producing. If everyone seems to be empowered to behave, then from what OpenAI experienced last yearwe are able to inform that issues might get uncontrolled.
Are CEOs prepared for that? If not, probably the most sure consequence of the continuing CEO AI psychosis will merely be organizational chaos.
Once you buy via hyperlinks in our articles, we may earn a small commission. This doesn’t have an effect on our editorial independence.
