A worrying trend is emerging in the corporate world, labelled 'AI psychosis' by Box founder Aaron Levie, where the individuals making decisions about AI replacing human roles often possess the least understanding of the intricacies of those very jobs. This disconnect, Levie suggests, is leading to a potentially misguided and accelerated push towards automation, with significant consequences for employment.
The impact of this phenomenon is already tangible. For instance, workplace productivity software firm ClickUp recently announced a substantial reduction in its workforce, cutting 22% of its employees. The company cited the integration of AI agents as a key factor in this decision, indicating a belief that these technologies can effectively take over a significant portion of human tasks.
This is not an isolated incident. Data reveals a concerning acceleration in tech sector layoffs. The number of job losses in 2026 due to technological shifts is already approaching the total figures recorded for the entirety of 2025. This rapid pace suggests a growing confidence, or perhaps overconfidence, in AI's immediate capabilities to streamline operations and reduce human capital, often without a full appreciation of the nuances involved in human-centric roles.
The core of the 'AI psychosis' argument rests on the idea that senior leadership, often far removed from the day-to-day operations and specific skill sets required for particular jobs, may have an idealised or oversimplified view of what those roles entail. This can lead to an underestimation of the human elements – such as critical thinking, emotional intelligence, complex problem-solving, and interpersonal communication – that are difficult for current AI models to replicate effectively.
As companies continue to explore and implement AI solutions, the challenge will be to bridge this understanding gap. A more nuanced approach, involving those directly performing the roles in question, could lead to more effective AI integration that complements human effort rather than simply attempting to replace it wholesale. Without this, the long-term implications for workforce stability and the quality of output remain uncertain.