Stable Diffusion, Ghibli LORAS: Steampunk industrial revolution factory interior, brass, copper, pipes, steam, gears, energy

If you want to predict the impact of a technology, try to imagine what will be made cheaper

Powerful technologies make lots of other things cheaper. For example, electricity made illumination, machinery operation, heating, and cooling cheaper, enabling a massive number of further inventions and productivity increases.

This is an interesting exercise to apply to AI: what do image generators and LLMs make cheaper? If data provenancing is improved, “search” would be one answer. Imagine using an image or test as a finger protocol to artists or experts!

A less positive reply would be “bullshit”: we can already see AI being used to deaden the internet with mindless SEO-optimized copy and clog teachers’ inboxes with stiffly-written 5-paragraph essays. A more precise response might be “the appearance of attention”. People use AI to simulate caring about stuff, like emails to subordinates, or assignments.

I don’t know how sustainable this huge and ongoing discount on giving a rat’s a** will be, but it does suggest that the value of genuine attention will rise.

“You can see the computer age everywhere but in the productivity statistics” - Robert Solow