People often discuss the amount of electricity required to train an AI model. But we almost never compare it to the energy required to “train” a human. Sam Altman recently made that comparison. Becoming an intelligent human takes roughly 20 years — the food consumed over a lifetime, education systems, buildings, teachers, transportation, and the global infrastructure that supports learning. Not to mention centuries of accumulated knowledge that every generation builds on. So the real comparison isn’t: AI training vs. a human answering one question. It’s: A fully trained AI model answering a question vs. A fully trained human brain doing the same task. Once models are deployed, the energy per query may already be reaching similar — or in some cases lower — levels of consumption. This shifts the conversation from “AI uses too much energy.” to how intelligence — artificial or biological — actually scales. We’re shifting from a debate over training costs to one about efficiency. --...