Artificial-Intelligence

AI capabilities are accelerating faster than most realize, driven by a fundamental shift in training methods. Reinforcement learning produced emergent reasoning nobody programmed. Fluid reasoning now matches average human performance. Autonomous task completion improves exponentially. The bubble arguments don’t hold: adoption is real, productivity gains are measurable, and demand outpaces supply.

AI has democratized programming by effectively adding 25 IQ points to developers’ capabilities, with less competent programmers seeing the largest productivity gains. This reduced barrier to entry, combined with a surge in computer science graduates (credentialism), has led to widespread under-employment.