Karpathy Says Developers Have ‘AI Psychosis.’ Everyone Else Is Next.

Why It Matters

The shift signals that AI‑driven productivity gains will soon affect every business function, while the accompanying job displacement and generational backlash could reshape talent strategies and risk management across industries.

Key Takeaways

• Developers experience “AI Psychosis” as frontier models solve code instantly
• Anthropic’s Claude Cowork expands agentic AI from coding to enterprise tasks
• Gallup study shows Gen Z AI anxiety rising, excitement falling sharply
• Goldman Sachs estimates AI cuts ~16,000 U.S. jobs monthly
• Anthropic’s Mythos model autonomously discovers zero‑day exploits, sparking security response

Pulse Analysis

The term "AI Psychosis" coined by Andrej Karpathy captures a moment where developers, armed with cutting‑edge models like OpenAI Codex and Anthropic Claude Code, see their traditional coding timelines collapse. This convergence of high‑performance AI, deep technical expertise, and revenue‑generating B2B software creates a preview of how agentic tools will reshape work across sectors. As these models become more reliable and integrated, they are no longer niche utilities for programmers but foundational platforms that can automate complex reasoning tasks in finance, law, and operations.

Enterprise adoption is accelerating. Anthropic’s recent launch of Claude Cowork moves the technology from a research preview into a full‑scale, plug‑in‑rich service aimed at non‑technical teams. By offering connectors to Google Drive, Gmail, DocuSign and industry‑specific applications, the company signals that the same productivity surge developers enjoy will soon be available to HR, marketing and supply‑chain professionals. This rollout mirrors the earlier diffusion of Claude Code, which took ten months to achieve mainstream traction, suggesting a similar adoption curve for broader business functions.

The broader impact extends beyond efficiency. A Gallup poll reveals Gen Z’s growing frustration with AI—anger up 9 points, excitement down 14—while Goldman Sachs estimates AI eliminates about 16,000 U.S. jobs each month, offset only partially by augmentation roles. Meanwhile, Anthropic’s Mythos model, which independently uncovered dozens of zero‑day vulnerabilities, highlights how rapid model improvements can generate unforeseen security challenges. Companies must therefore balance the competitive advantage of aggressive AI token consumption—exemplified by Meta’s internal leaderboard—with robust governance to mitigate talent displacement and emerging cyber risks. The coming months will test how quickly organizations can adapt policies, upskill workforces, and secure AI‑driven workflows.