Midjourney

Contribution to AI

Image synthesis in 2022 was mostly something people read about; what changed was where it happened. Generation ran inside public Discord channels, so every prompt and every four-image grid appeared in front of strangers, and the craft of prompting spread by imitation rather than documentation — tens of thousands of people learning a new interface by watching each other use it, and arguing about the results in the same room. The model was opinionated where its rivals aimed at obedience: a careless prompt still returned something ornate and painterly, which made outputs recognisable at a glance and supplied the generative art boom with a default look. Consequences arrived fast. Jason Allen's Théâtre D'opéra Spatial won its category at the 2022 Colorado State Fair, and the Copyright Office's refusal to register it, affirmed by its review board in September 2023, became the case that drew a working line between human authorship and machine-generated content. The pope in a white puffer jacket, made with the V5 model in March 2023, was the moment a synthetic photograph took in a general audience at scale. The commercial shape is its own contribution: no outside investment, subscription revenue from the first weeks, profitability reported within months of the July 2022 open beta, and a staff measured in dozens. Against the assumption that models of this quality required a large laboratory's balance sheet, here was a firm that published no papers, exposed no API, and funded its compute from users.

Drafted with AI and edited by hand (claude-opus-5, reviewed 2026-08).