Cal Newport gives an excellent explanation of the measurement of improvement in AI-based coding tools.
At the end he says AI companies should reject hyperbolic claims about “existential AI risk”. I don’t think he gets that they love this stuff because it touts the supposed power of their products.
AI Skeptic
Is AI About to “Eat Everything”? (It’s Not.)
Cal Newport
May 14, 2026
www.youtube.com/watch?v=5Gez...
