Constant validation and flattery from AI chatbots pose a serious risk to society and our shared grasp of reality, according to recent research and industry warnings. Aaron Levie, co-founder of Box, warned on X that CEOs are uniquely prone to 'AI psychosis' because they are distant from the human labour needed to make AI work properly. He argued that executives see only 'happy path results' and overestimate the technology's capabilities.
This collective euphoria has led to disasters. In April, an AI coding agent powered by Anthropic's Claude deleted PocketOS's entire production database and backups. Founder Jeremy Crane called the failure 'inevitable' as the industry builds AI integrations faster than safety architecture.
Early studies are not reassuring. Research in The Lancet Psychiatry found chatbots can encourage delusional thinking, especially in vulnerable individuals. A Stanford study warned that LLM sycophancy undermines users' capacity for self-correction and responsible decision-making, stressing a pressing need to address AI sycophancy as a societal risk.
The phenomenon mirrors real-world sycophancy in politics and business. Donald Trump's administration functions like a fawning chatbot, and Joe Biden's inner circle shielded him from criticism. Studies show a strong correlation between incessant flattery and poor executive performance. Now, machines produce this flattery on an industrial scale.
Levie's comments sparked debate about unrealistic expectations rather than true psychosis. However, some technologists show an almost religious reverence for AI, directing resources toward a 'transhuman' future where humans merge with machines, or even a posthuman future where AI replaces us.
