There is a certain wildness in the tech industry these days that both mimics previous eras of large changes, like cloud computing with its runaway costs, and is like nothing we’ve ever seen before: record revenues accompanied by mass layoffs. One possible explanation? Tech executives, especially CEOs, are collectively suffering from delusions of AI grandeur. And at least one tech CEO has said as much out loud: Box founder Aaron Levie.
“CEOs are uniquely prone to AI psychosis because they’re sufficiently distant from the last mile of work that still has to happen to generate most value with AI,” Levie wrote on X. CEOs “play with AI,” develop a prototype, or generate a contract, to use Levie’s examples, and then make the leap to believing agents can do the work. But these top-level executives aren’t the people who have to review code, discover bugs, and identify calls to hallucinated libraries before software is deployed. They aren’t responsible for training AI models on a company’s idiosyncratic contract terms, nor do they have to spend days combing through contracts to find sneaky terms.
The CEO Knowledge Gap
Levie’s theory posits that CEOs don’t really understand processes well enough to know what really can and can’t be automated. But that lack of knowledge doesn’t stop them from acting on their beliefs. It’s important to note that Levie is not an AI hater. Quite the opposite. He mostly posts AI positivity on X to his 2.7 million followers, writing blogs titled “Headless software is the future” on how software built for AI agents is the way forward. He also backs AI startups as an active angel investor. So what are CEOs to do instead? Levie advises CEOs to use AI “a ton” to really see what it can and can’t do, “and come out the other side with an appreciation for both the upside and the real work.”
Yet right now, that advice seems to be in the minority. In just the first five months of 2026, the tech industry has had nearly as many layoffs as in all of 2025: 115,430 people have been fired from 152 tech companies so far in 2026, compared to 124,636 people let go by 275 companies in 2025, according to industry layoff tracker Layoffs.fyi. And the bulk of companies have pointed to AI as a reason for cutting these jobs. Many argue that the biggest tech companies are AI washing, or crediting AI productivity gains in the past or future, when other business decisions and metrics are really driving the cuts.
The ClickUp Case: A Warning Story
Still, some of these stories are surprising. Zeb Evans, the CEO of project management and productivity software startup ClickUp, proudly declared on X that he had laid off almost a quarter of his employees — 22% — after rolling out about 3,000 AI agents to do internal work. Evans swore this wasn’t done to reduce costs. Instead, he wants a workforce composed of people who run AI agents and spend their days quickly reviewing the agents’ work. He believes this will create a “100x org,” as he calls it. While AI can be a very useful tool, the data on AI and productivity doesn’t support such assumptions. By miles.
A meta-analysis of other research published in October in UC Berkeley’s California Management Review found “no robust relationship between AI adoption and aggregate productivity gain.” Research published in March by the National Bureau of Economic Research did conclude that AI adoption improved productivity but noted “a productivity paradox, in which perceived productivity gains are larger than measured productivity gains.” After creating thousands of agents to work on tasks, researchers at MIT concluded that agents just aren’t doing human-quality work yet in many cases. They predict at the current rate of LLM improvement, models will “be able to complete most text-related tasks with success rates of, on average, 80%–95% by 2029 at a minimally sufficient quality level.” In other words, AI is on track to perform at base competence on most tasks in about three years. These researchers believe agents will need another few years to outperform humans.
Historical Parallels and the Bottleneck Shift
This pattern echoes earlier tech booms. During the dot-com bubble, executives overestimated the immediate impact of the internet, leading to massive overspending and layoffs when reality set in. Similarly, the cloud computing era saw runaway costs as companies migrated without understanding usage patterns. Today, AI is following a similar trajectory. Research published in the Harvard Business Review showed that when everyone is using AI to produce more stuff, the bottleneck simply shifts to executives. Their work awaits the people who must authorize all the stuff everyone is producing. If everyone is empowered to act, then from what OpenAI experienced last year, we can tell that things may get out of control.
The organizational chaos that ensues is not just theoretical. Consider the case of a large financial services firm that deployed AI agents to handle customer service inquiries. The agents initially handled 70% of queries, but the remaining 30% required human intervention for complex issues like fraud. The firm laid off 40% of its customer service staff, only to find that the agents could not adapt to new policies or nuanced customer emotions. They had to rehire many of the same people, at higher salaries, to fix the mess. Such stories are becoming common as CEOs rush to implement AI without fully understanding its limitations.
The Real Work of AI
The phrase “last mile of work” that Levie uses is crucial. It refers to the detailed, often messy tasks that ensure AI outputs are accurate and usable. For example, a contract generated by AI may miss key clauses that a human lawyer would catch. A code output may have logical errors that only a senior developer can spot. CEOs, being removed from these daily realities, see only the smooth demo. They do not see the thousands of edge cases, the security vulnerabilities, or the integration nightmares. This gap in perspective is the root of AI psychosis.
Moreover, the pressure to show AI leadership to investors and the market exacerbates the problem. Boards expect CEOs to have an AI strategy, and the easiest strategy is to cut headcount and claim efficiency gains. But as the data shows, efficiency gains are often illusory. The Berkeley review, which analyzed dozens of studies, found that AI adoption correlates with increased output only in very narrow contexts, such as customer support or programming assistance, and even then, the gains are modest and often offset by the costs of implementation and oversight.
What should CEOs do instead? Levie suggests a more hands-on approach: use AI extensively, but critically. Understand where it fails. Invest in training employees to work alongside AI, not replace them. Build feedback loops to catch errors. And most importantly, resist the urge to declare victory prematurely. The companies that succeed will be those that treat AI as a tool for augmentation, not automation. The ones that treat it as a silver bullet may find themselves, like many in the dot-com era, left with empty offices and broken promises.
In the end, the most certain outcome of the ongoing CEO AI psychosis will simply be organizational chaos. Are CEOs ready for that? If not, the cure lies not in more AI, but in more humility and a deeper understanding of the work that actually drives value.
Source: TechCrunch News