There is a certain wildness in the tech industry these days that both mimics previous eras of large changes—like cloud computing's runaway costs—and is like nothing we've ever seen before: record revenues accompanied by mass layoffs. One possible explanation, gaining traction among industry observers, is that 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.

Levie, a well-known figure in enterprise software and an active angel investor in AI startups, recently took to X (formerly Twitter) to coin the term "AI psychosis." In a post that garnered widespread attention, he wrote: "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." The remark struck a chord because it articulated a frustration many engineers and product managers have felt privately: that top-level executives, playing with shiny prototypes, leap to the conclusion that AI agents can fully automate complex workflows—without understanding the messy reality of implementation.

Levie's point is that CEOs see the happy path results of AI—a generated contract, a prototype chatbot, a quick code snippet—but they rarely consider the next 10 or 20 things that have to happen to turn that into a reliable, production-grade system. They aren't the ones 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 or spending days combing through fine print to spot sneaky clauses. In other words, Levie's theory posits, CEOs don't really understand processes well enough to know what can and can't be automated. But that lack of knowledge doesn't stop them from acting on their beliefs.

The Layoff Paradox

This disconnect has real-world consequences. In just the first five months of 2026, the tech industry has seen 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 those companies have pointed to AI as a reason for cutting jobs. Critics argue that many of these layoffs are merely "AI washing"—crediting productivity gains that haven't materialized—while the real drivers are cost-cutting, shareholder pressure, or overhiring during the pandemic boom.

Nevertheless, some CEOs have been remarkably open about their faith in AI as a replacement for human workers. Zeb Evans, CEO of project management software startup ClickUp, proudly declared on X that he laid off nearly a quarter of his employees—22 percent—after rolling out about 3,000 AI agents to handle internal work. Evans insisted the layoffs were not about reducing costs but about creating a "100x org" staffed by a small number of people who run AI agents and review their output. His announcement sparked heated debate, with many noting that such a drastic move seemed to ignore the ample research showing that AI, while useful, is far from ready to replace humans in most roles.

What the Research Says

The data on AI and productivity does not support the assumption that deploying AI agents leads to proportional workforce reduction. A meta-analysis of research published in October 2025 in UC Berkeley's California Management Review found "no robust relationship between AI adoption and aggregate productivity gain." A March 2026 study by the National Bureau of Economic Research concluded that AI adoption improved productivity but noted "a productivity paradox, in which perceived productivity gains are larger than measured productivity gains." Meanwhile, researchers at MIT, after running extensive experiments with AI agents, predicted that current large language models (LLMs) would likely achieve sufficient quality for most text-related tasks by 2029—but only at a minimally acceptable level. They estimate it will take another few years for AI to outperform humans consistently.

That timeline clashes with the aggressive deployment schedules many CEOs are now pursuing. ClickUp's Evans, for instance, expects his 3,000 AI agents to already be performing at or above human levels. But the MIT researchers caution that even in 2029, agents will still require significant human oversight. In many complex, multi-step tasks, failure rates remain high. Agents hallucinate, misunderstand context, and struggle with domain-specific terminology. The so-called "last mile" problem—the gap between a prototype and a reliable, enterprise-grade solution—remains the industry's biggest challenge.

Historical Parallels

The current AI frenzy bears striking resemblance to earlier technology hype cycles. During the cloud computing boom of the early 2010s, many companies rushed to migrate workloads expecting immediate cost savings, only to face ballooning expenses from poorly managed cloud instances. Similarly, during the dot-com bubble, executives convinced themselves that the internet would instantly transform all business models—leading to massive overspending and subsequent crashes. The difference this time is that AI is being used to justify not just investment but also large-scale human displacement, which carries serious social and ethical implications.

Furthermore, the dynamics of AI adoption create new bottlenecks. Research published in the Harvard Business Review showed that when everyone in an organization uses AI to produce more output, the bottleneck shifts to executives. They become overwhelmed by the volume of generated content, proposals, and decisions that require authorization. Unless companies redesign their workflows and decision-making hierarchies, the increased output from AI could simply result in organizational chaos. This is not a hypothetical fear; last year, OpenAI itself experienced internal governance challenges when its AI tools enabled employees to produce and approve actions at a rate that outpaced oversight, leading to costly errors.

Is There a Cure for AI Psychosis?

Levie, despite his diagnosis, remains a strong advocate for AI. He posts mostly AI-positive content to his 2.7 million followers, and he has written blogs such as "Headless software is the future," arguing that software built specifically for AI agents is the way forward. He also puts his money where his mouth is, backing numerous AI startups as an angel investor. His advice to CEOs is not to abandon AI but to use it "a ton"—to deeply understand its capabilities and limitations through hands-on experimentation, rather than relying on demos and sales pitches. "Come out the other side with an appreciation for both the upside and the real work," he wrote.

Unfortunately, many CEOs appear to be in a hurry. The pressure to show AI-driven results—to investors, boards, and the market—often overrides the need for careful empirical testing. Moreover, the competitive landscape rewards first movers, even if their moves are premature. As a result, the tech industry may be in for a period of significant organizational turbulence. Workers are being let go en masse, but the promised productivity gains have yet to materialize. If history is any guide, the companies that survive and thrive will be those that strike a balance between ambition and realism, blending human expertise with machine assistance rather than attempting wholesale replacement.

The story of tech's AI psychosis is still unfolding, but it serves as a cautionary tale about the dangers of executive overconfidence. The gap between what AI can do in a demo and what it can do in the real world is the source of many bad decisions. Until CEOs close that gap by getting their hands dirty with the actual work, the layoffs will continue, the confusion will persist, and the promise of AI will remain partial at best.

Source: TechCrunch News