In the tech industry, a peculiar wave of executive overconfidence is sweeping through corner offices. It's a phenomenon that Box founder and CEO Aaron Levie has labeled "AI psychosis": a state where tech CEOs, seduced by the promise of artificial intelligence, believe that AI agents can handle the most complex, last-mile work of their organizations—without understanding the gritty details required to make it happen. This delusion is leading to sweeping layoffs, unrealistic productivity claims, and potentially chaos within companies.
Levie, speaking on X, explained that 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." When executives play with a prototype or generate a contract using AI, they see only the happy path—the successful result. They do not see the dozens of subsequent tasks that must be performed manually: reviewing code for bugs, catching calls to hallucinated libraries, training models on idiosyncratic company data, or combing through contracts for sneaky terms.
The disconnect between CEO vision and ground reality
This disconnect is not born from a lack of enthusiasm for AI. Far from it. Levie himself is an AI optimist, actively backing startups and writing about how "headless software is the future." But even he recognizes that CEOs often fail to appreciate what truly can and cannot be automated. The problem is that this lack of deep knowledge does not prevent them from acting on their beliefs. They announce massive workforce reductions, claiming AI will fill the gaps, and then leave middle managers and engineers to figure out the impossible.
One striking example is Zeb Evans, CEO of project management software startup ClickUp. In a public declaration on X, Evans revealed that he had laid off 22% of his workforce—almost a quarter of the company—after deploying about 3,000 AI agents to handle internal tasks. Evans insisted this was not a cost-cutting measure. Instead, he envisions a future workforce composed of people who run AI agents and review their output, creating what he calls a "100x org." The underlying assumption is that AI agents are already competent enough to replace human workers in a wide range of duties, and that humans only need to supervise them.
Yet the data tells a different story. A meta-analysis published in October in the California Management Review from UC Berkeley found "no robust relationship between AI adoption and aggregate productivity gain." Similarly, research from the National Bureau of Economic Research, published in March, concluded that while AI adoption improved productivity, there was a "productivity paradox" in which perceived gains are larger than measured gains. The gap between what executives believe AI can deliver and what it actually delivers is widening.
The human cost of AI psychosis
The most visible consequence of this mindset is the wave of tech layoffs. According to industry tracker Layoffs.fyi, in just the first five months of 2026, 115,430 people have been fired from 152 tech companies. That number is nearly as high as the entire 2025 total of 124,636 people let go from 275 companies. The majority of these companies have cited AI as a reason for the cuts, though critics argue that many are "AI washing"—using AI hype to mask other business decisions or metrics that are actually driving the reductions.
Regardless of the motivation, the impact on employees is severe. Jobs are being eliminated not because the work has disappeared, but because executives believe—often incorrectly—that AI can handle it. The MIT researchers who studied the capabilities of large language models found that at the current rate of improvement, AI will only be able to complete most text-related tasks with success rates of 80% to 95% by 2029 at a "minimally sufficient quality level." That is still three years away, and even then, the models will not outperform humans. The researchers estimate it will take several more years for AI to exceed human performance. So CEOs who are firing people today in favor of AI are betting on a future that has not arrived.
The bottleneck shifts to executives
Further complicating the picture is research published in the Harvard Business Review, which found that when everyone uses AI to produce more output, the bottleneck simply moves upward. Executives become the choke point because they must authorize the increased volume of work generated by AI-augmented teams. If every employee is empowered to act, the organization can spiral out of control—as OpenAI experienced last year when its AI-powered tools led to unexpected consequences and overspending.
In other words, even if AI could do everything CEOs claim, the next crisis would be executive capacity. But since AI cannot yet deliver on those promises, companies are left with fewer human workers doing the same work, often with less support. The result is burnout, lower-quality outputs, and missed deadlines.
Levie's advice to CEOs is to use AI "a ton"—not to replace workers, but to genuinely understand what the technology can and cannot do. Only by immersing themselves in the granular reality of AI's limitations can CEOs come out the other side with an appreciation for both the upside and the real work that remains. But right now, too many executives are skipping that step. They see a demo, imagine a future, and act on that vision without rigorous testing.
The tech industry has seen similar cycles before. In the early days of cloud computing, companies rushed to migrate and were shocked by runaway costs. But the current AI boom is different: it is happening at a time of record revenues for many tech giants, yet those records coincide with mass layoffs and a palpable anxiety among employees. The combination of financial success and workforce reduction is unprecedented, and the central driver is C-suite belief in AI's transformative power.
Until CEOs confront the data and the real-world limitations of AI, the industry will continue to experience "AI psychosis"—a collective delusion that leads to poor decisions, wasted resources, and unnecessary human suffering. The most likely outcome, if this trend continues, is not a utopia of 100x organizations but organizational chaos.
Source: TechCrunch News