In the rapidly evolving tech landscape, a growing concern is emerging: many CEOs are suffering from AI psychosis, a term coined by Box founder Aaron Levie. Levie argues that top executives are too removed from the actual work required to fully leverage AI, leading to unrealistic expectations. "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. He noted that CEOs often experiment with AI tools, creating prototypes or generating contracts, and then mistakenly believe these can replace human work. However, these executives aren’t involved in the detailed tasks like reviewing code, identifying bugs, or training models on specific contract terms.

They also don’t have to spend days analyzing complex contracts, as Levie explains. Despite this, Levie isn’t anti-AI. He actively supports AI startups as an angel investor and promotes AI-driven software solutions. The issue, he suggests, is that CEOs need to engage more with AI to understand its true capabilities and limitations. Meanwhile, the tech industry has seen a significant rise in layoffs, with 115,430 employees laid off in the first five months of 2026, according to Layoffs.fyi.

Many companies cite AI as a reason for these cuts, though some argue it's more about cost-cutting than genuine productivity gains. For instance, Zeb Evans, CEO of ClickUp, laid off 22% of his workforce after implementing 3,000 AI agents, claiming it would create a "100x org." However, research indicates that AI adoption hasn't significantly boosted productivity. A meta-analysis published in October by UC Berkeley found no robust link between AI adoption and productivity gains. Similarly, a March study by the National Bureau of Economic Research noted a productivity paradox, where perceived gains exceed actual measurements.

MIT researchers also found that current AI agents don’t match human performance in many tasks, predicting they’ll reach 80%-95% success rates by 2029. As AI capabilities grow, the responsibility for managing outputs may shift to executives, who could face organizational chaos if unprepared.

Source: techcrunch