Abstraction

I've noticed a few tropes in the images people place at the top of blog posts about AI: connected nodes, boxes, brains, even humanoid robots. And always blue.

These tropes feel lazy, sloppy, or thoughtless. They also project ideas that LLMs are mystical or magical, existing in some virtual world abstracted from reality.

Large Language Models require real infrastructure, real materials, and real labor to operate in the real world. In my opinion, using images of that infrastructure paints a more honest picture of what "AI" actually is than throwing up some blue boxes.

Please consider un-abstracting "AI".