Curve Caution
Sam Altman lays out a clean AI timeline: novel insights by 2026, robot labor by 2027. But tech rarely cooperates. In the 1970s, symbolic AI seemed inevitable. In the 1990s, neural nets were dismissed. In 2000, few saw smartphones changing everything. Real breakthroughs usually come sideways. Today’s models are powerful, but we already see improvements slowing. Clean data is running out. Hallucinations persist. New releases feel like tuning, not leaps. Progress now depends on tools, memory, and synthetic feedback. Altman’s optimism is welcome, and his openness matters. But as Thomas Kuhn argued, science moves through paradigm shifts—and this one may have already happened when ChatGPT arrived two years ago. Read his blog post →