Aaron Levie, CEO of the cloud platform Box, has reported that an "AI psychosis" is spreading among leaders in the tech world. According to him, many top managers are overestimating the real capabilities of AI systems and drawing hasty conclusions about labor automation. Levie noted that executives often see only demo versions of the technology and assume that agents can perform complex processes without human intervention. This is reported by Ixbt.com reports.
In reality, leaders have become detached from daily practical workflows. They do not look for bugs in code, nor do they check for model "hallucinations"—data and links invented by AI. Levie is not against AI; he even actively invests in the field, but he urges leaders not to rely solely on presentations and to deeply understand the limitations of the technology.
The situation is further exacerbated by mass layoffs in the IT sector. According to Layoffs.fyi, more than 115,000 employees have lost their jobs in the first five months of 2026. For example, the service ClickUp laid off 22% of its staff after implementing 3,000 AI agents. Company CEO Zeb Evans called this not a cost-cutting measure, but the creation of a new type of organization where employees only verify AI results.
However, scientific research has yet to confirm such drastic efficiency. Analyses by the California Management Review and the National Bureau of Economic Research (NBER) have not found a stable correlation between AI implementation and overall productivity. According to MIT researchers, large language models may only be able to perform text-based tasks at a human level by 2029, but achieving full superiority will take several more years.
Read “Zamin” on Telegram!
