AI psychosis is emerging as a worrying phenomenon among the most intensive users of artificial intelligence tools, particularly developers who rely on AI agents for coding. Andrej Karpathy, co-founder of OpenAI and the man who coined the term "vibe coding," has described his own state as "AI psychosis" since December.
In an interview with Sarah Guo, Karpathy revealed that he works 16-hour days directing swarms of code agents. He also admitted to feeling "extremely nervous" when he has unused tokens left at the end of the month—a behavioral signature of someone who has internalized scarcity as a threat.
What Is AI Psychosis and Why Is It Trending?
AI psychosis refers to a pattern of behavioral deterioration observed in heavy users of AI tools. It's not a clinical diagnosis but a colloquial term capturing symptoms like obsessive usage, anxiety over resource limits, and disrupted sleep. The term has gained traction as more high-profile tech figures share similar experiences.
Karpathy's case is not isolated. Garry Tan, CEO of Y Combinator, has called his own experience "cyber psychosis." A CTO interviewed by Axios said he needs prescription medication to sleep. These are not anecdotal profiles; they represent a growing trend among the most productive users of AI.
Why AI Psychosis Matters for the Future of Work
The dominant narrative around AI has been one of limitless productivity and the famous "10x" developer. However, what's now being documented is the dark reverse: the most intensive users are also the ones showing the most concerning signs of behavioral deterioration.
If the most productive tools in history generate the same patterns in their heaviest users as gambling, the debate on AI's impact on work enters another dimension. It raises questions about sustainability, mental health, and the true cost of hyper-productivity.
Key Takeaways
- AI psychosis describes behavioral deterioration in heavy AI users, including anxiety over unused tokens and obsessive work patterns.
- High-profile figures like Andrej Karpathy and Garry Tan have publicly shared their experiences with the phenomenon.
- The most intensive AI users may exhibit signs similar to gambling addiction, such as withdrawal anxiety and sleep disruption.
- This trend challenges the narrative of limitless productivity and raises concerns about mental health in tech.
- Developers and companies need to watch for warning signs and promote healthier AI usage habits.
How AI Psychosis Manifests: Signs and Symptoms
AI psychosis can manifest in various ways. Common signs include working excessively long hours, feeling anxious when not using AI tools, and neglecting sleep and personal care. Karpathy's 16-hour days and nervousness over unused tokens are prime examples.
Other reported symptoms include a sense of losing control over one's work habits, social withdrawal, and an inability to disconnect. These patterns mirror those seen in behavioral addictions, where the reward system is hijacked by the activity.
Comparing AI Psychosis to Other Tech-Related Conditions
AI psychosis shares similarities with other tech-induced conditions like burnout and gaming addiction. However, its unique trigger—the use of AI agents—sets it apart. The table below compares key aspects. The background to this story is set out in Black Box Podcast Season 2 Explores AI Chatbots and Psychosis. A related thread runs through Elon Musk's Cyber Psychosis: What AI Users Need to Know. Earlier coverage in AI and Developer Mental Health: The Identity Crisis in Tech shows how the argument developed.
| Condition | Primary Trigger | Common Symptoms | At-Risk Group |
|---|---|---|---|
| AI Psychosis | Heavy use of AI agents | Obsessive work, token anxiety, sleep disruption | Developers, tech leaders |
| Burnout | Chronic workplace stress | Exhaustion, cynicism, reduced efficacy | All professions |
| Gaming Addiction | Video games | Preoccupation, withdrawal, tolerance | Adolescents, young adults |
What to Watch Next
As AI tools become more integrated into work, the phenomenon of AI psychosis is likely to grow. Companies and individuals should monitor usage patterns and promote healthy boundaries. The conversation is shifting from pure productivity gains to sustainable practices. A closer look at the same subject appears in Ai Video: latest articles. Anyone following the wider picture can start with Google's AI Now Generates 75% of New Code: What It Means. The same theme surfaces in AI Relationship Advice: Why Claude's Flattery Can Harm You, which takes a different angle.
Karpathy's openness about his experience is a wake-up call. It's essential to recognize the signs early and seek balance. The future of work with AI depends on how we manage its psychological impact. Context on how this played out before sits in Understanding Postpartum Psychosis: Key Facts and Support. A companion piece, Lindsay Clancy Trial and Postpartum Psychosis Awareness, approaches the same subject from another angle.