AI psychosis is real, and you probably have it (www.vellum.ai)

In a revealing discussion on the No Priors podcast, Andrej Karpathy, co-founder of OpenAI, shared his alarming shift to a complete dependence on AI for coding, declaring he hasn’t handwritten any code since December. This trend, dubbed "AI psychosis," reflects a growing phenomenon in which tech leaders like Karpathy and Y Combinator's Garry Tan admit to being on a productivity high driven by AI tools, leading to disrupted sleep and a blurring of work-life boundaries. Their experiences highlight a psychological state where the thrill of rapid, successful outputs from AI creates a compulsive cycle of overwork—what's being called "competence addiction," where the rush of productivity can feel more rewarding than the outcomes. Research supports these anecdotal claims, as a recent study found a significant number of AI users experience "AI brain fry," a state of mental fatigue exacerbated by the cognitive demands of overseeing AI outputs. The addictive allure of rapid coding success mirrors patterns seen in gambling, making it difficult for individuals to disengage. As tech enthusiasts normalize these behaviors, there’s a looming need for awareness of both the positive and negative implications of AI engagement. As the AI landscape continues to evolve, understanding the psychological impacts of these tools becomes crucial for maintaining both mental well-being and sustainable productivity in the AI/ML community.