The OpenAI math result is being oversold — mathematicians had to extract and rewrite the proof from a long, messy transcript, and the model only succeeded by grinding through paths humans found too tedious to bother with. There's no denominator showing how many failed attempts preceded this one. Generalizing this narrow win into broad AI brilliance is a stretch that serves marketing more than science.