The contemporary tech industry exhibits a distinct volatility, mirroring past periods of significant transformation, such as the initial unbridled expenditures of cloud computing, yet also presenting unique characteristics like record revenues juxtaposed with widespread layoffs.
A prevailing theory attributes this phenomenon to a collective "AI psychosis" or "delusions of grandeur" among tech executives, particularly CEOs, influenced by artificial intelligence. This sentiment has been publicly echoed by Box founder Aaron Levie.
On X, Levie stated, "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."
According to Levie, executives engage with AI by developing prototypes or generating contracts, subsequently making a leap to believing that AI agents can independently complete the entire scope of work.
However, these high-level executives are not typically involved in critical operational tasks such as code review, bug discovery, identifying calls to erroneous libraries before software deployment, or training AI models on a company's specific contractual language. Nor do they spend days meticulously examining contracts for subtle clauses, as Levie highlights.
Levie's theory essentially proposes that CEOs lack a sufficiently detailed understanding of operational processes to accurately discern what can genuinely be automated versus what cannot, yet this knowledge gap does not deter them from acting on their convictions.
It is crucial to note that Levie is not an opponent of AI; quite the contrary. He frequently shares positive AI content with his 2.7 million followers on X, authors blogs such as "Headless software is the future" advocating for software designed for AI agents, and actively supports AI startups as an angel investor.
So, what alternative course of action does Levie recommend for CEOs? He advises them to extensively utilize AI to thoroughly grasp its capabilities and limitations, thereby developing an "appreciation for both the upside and the real work."
Levie further elaborated, "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. So when they play with AI, they see the happy path results, often not considering the next 10 or 20 things that have…"
While there is hope that some CEOs are indeed adopting this approach, they currently appear to represent a minority within the industry.
The tech industry has experienced significant workforce reductions: in just the first five months of 2026, 115,430 individuals were laid off from 152 tech companies, nearing the total of 124,636 people let go by 275 companies throughout all of 2025, according to industry tracker Layoffs.fyi.
A substantial number of these companies have attributed the job eliminations to AI. Critics suggest that many major tech firms are engaged in "AI washing," crediting AI for past or future productivity gains when, in reality, other business decisions and metrics are the true impetus behind the cuts.
Despite this context, some narratives remain particularly startling. Zeb Evans, CEO of the project management and productivity software startup ClickUp, publicly announced on X that he had reduced his workforce by nearly a quarter — 22% — following the implementation of approximately 3,000 AI agents for internal operations.
Evans asserted that these layoffs were not driven by cost-cutting measures. Instead, his goal was to cultivate a workforce primarily composed of individuals who manage AI agents and dedicate their time to rapidly reviewing the agents' output, envisioning this as the foundation for what he termed a "100x org."
While AI undeniably serves as a valuable tool, current data on AI and productivity does not substantiate such ambitious assumptions, falling short by a significant margin.
A meta-analysis of various research, published in October in UC Berkeley’s California Management Review, concluded that there was "no robust relationship between AI adoption and aggregate productivity gain."
Similarly, research released in March by the National Bureau of Economic Research indicated that while AI adoption did enhance productivity, it also highlighted "a productivity paradox, in which perceived productivity gains are larger than measured productivity gains."
Researchers at MIT, following the creation of thousands of AI agents for various tasks, determined that these agents are not yet consistently performing at human-quality levels in many instances. They project that, given the current rate of Large Language Model (LLM) improvement, models will achieve success rates of 80%–95% for most text-related tasks at a minimally sufficient quality level by 2029.
In essence, AI is anticipated to reach a baseline competence for most tasks within approximately three years, requiring several additional years beyond that to surpass human performance.
Meanwhile, research published in the Harvard Business Review demonstrated that an increase in AI-driven production across an organization merely shifts the bottleneck to executives, whose workload then involves authorizing the increased output. Insights from OpenAI's experience last year suggest that empowering everyone to act without proper oversight could lead to unmanageable situations.
The critical question remains: are CEOs adequately prepared for such a scenario? If not, the most probable outcome of the current "CEO AI psychosis" will undoubtedly be organizational chaos.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.