[AI Summary]: A recent METR study found that experienced software developers completed tasks 20% slower when using AI tools, contradicting predictions of 40% productivity gains and challenging the narrative driving massive AI investments. Despite tech giants spending $560 billion on AI infrastructure while generating only $35 billion in AI revenue, and MIT research showing 95% of AI initiatives failing to boost profits, the U.S. economy is increasingly dependent on AI-driven valuations with the Magnificent Seven stocks accounting for over half of S&P 500 growth. This disconnect between AI’s promised productivity revolution and its real-world performance raises concerns about a potential bubble that could exceed the dot-com crash, potentially triggering recession or financial crisis if investor confidence collapses.
- Author: Rogé Karma
- Published: September 7, 2025
- Source: The Atlantic