Raw LLM Responses

**Summary** The piece presents a dystopian scenario explaining how rapid advances in AI could systematically replace human labor and reshape society, a phenomenon it calls the **“intelligence curse.”** It begins with companies adopting AI agents that can do junior-level work faster and cheaper than humans. Competitive pressure forces even reluctant firms to replace workers. Junior employees are laid off first, then managers, then executives, as AI proves better at coordination, optimization, and decision-making. Productivity and stock markets soar, but unemployment rises to unprecedented levels, surpassing the Great Depression. Efforts to respond through policies like Universal Basic Income fail due to political resistance and budget concerns, while protests are muted by media and platform control. Wealth and power increasingly concentrate in AI firms and their shareholders, who become the government’s main constituency. As human labor loses value, public institutions like schools deteriorate, upward mobility disappears, and society regresses into a rigid elite structure where wealth is inherited rather than earned. The author argues this outcome is plausible if: * AI continues improving beyond human intelligence, * AI becomes cheaper than human labor, * Companies prioritize profit over employment, * Competitive pressure forces universal adoption, * Media and politics align with corporate interests. This dynamic mirrors the **resource curse**, where countries rich in oil or minerals become poorer for citizens because wealth concentrates among elites. AI, as a form of “intelligence extraction,” could do the same globally—eliminating human bargaining power entirely. The piece concludes by arguing that not all AI is dangerous. The real risk comes from **general, autonomous intelligence** that can do nearly everything humans can. Narrow or specialized AI (like protein-folding systems or self-driving cars) is beneficial, but general AI threatens mass displacement and potentially runaway self-improvement. Preventing this future requires careful limits on how intelligent, autonomous, and general AI systems are allowed to become.