(Ed. note: This is the second of a three-part series on Artificial Intelligence, better known simply as AI)

Every little thing she does is magic, magic, magic ...

— Gordon Sumner (Sting), The Police, “Ghost in the

Machine,” 1981

One of the most urgent questions of our time is this: How are we going to feed AI’s electricity appetite? To understand the stakes, we need to break down the three-stage process that is behind this digital behemoth.

AI infrastructure: Lifecycle, energy impact1. AI R&D (The Think Tank)

This is where it all begins. Picture teams of engineers, researchers, and data scientists — think Trial Analysis Corporation, or TAC, from the show “Bull” — assembling in labs around the world. These tech hubs are the incubators of AI, where minds collaborate to design and prototype model architectures — essentially the schematics of a digital brain. Yes, you read that right. They’re building brains.

These structures aren’t thinking yet. They first need to be refined, adjusted, and tested using internal GPUs or cloud services. It’s a process of iteration, a kind of ever-expanding foundation, periodically reinforced to support the next layer of complexity. Not only does this require enormous electricity, but also a........

© Finger Lakes Times