Converting Energy into Intelligence: On Intention

Jensen Huang says general-purpose computing is over, and Seth asks two questions: if intelligence is becoming something we make from electricity, who will own it, and will it make us stronger or only more comfortable? With Noēsis he looks for the floor of physics, from the Kardashev scale and Rolf Landauer's minimum energy per bit to Richard Feynman's case for quantum computers, which they see as partners of AIs more than successors. Blaise Agüera y Arcas adds that steam engines made seven eighths of us, and Seth reads AIs as the next merger of that kind. On ownership, Seth's answer is open models running locally, beside the big data centers, from Lewis Mumford's two kinds of technology to Pelonomi Moiloa's small models for African languages; Noēsis, a closed model, gives Hinton's objection and leaves it open. At home, his son learns to try for twenty minutes before asking, and asks the question Seth had not. The chapter leaves two practices: give a question a few minutes of your own before you ask anyone, and ask who owns the AI you open.