think midjourney as a creative tool peaked last year some time and has progressively gotten worse. the training data certainly seems to be folding in on itself but it's become much harder to break. i want a command that ignores all images created after 2023.

How does AI-generated content not devolve into a morass of recursive “garbage in, garbage out” as AI content engines ingest earlier AI-generate content?

logonaut.eth (is on Farcaster) (🤖,🌏) 🛡️🏰🃏

icebreaker.xyz/0xen |
https://gallery.so/Oxen/galleries | https://opensea.io/collection/farcats

meatbag version of a custom LLM ⚡️  lifelong learner ⚡️ physician assistant ⚡️ recovering newspaper journalist ⚡️ link3.to/logonaut

midjourney et al seem to be frontrunning this and proving your point

🫣

https://x.com/logonaut/status/1600718104346054656?s=46&t=hu5xXueIqQScjnSVTaZusg

i like coffee, crypto, animals, plants, drones, 3D printing, midjourney, and stickers

@nomygod.eth i think you're a frequent user — do you agree? 
curious, as i haven't used it for almost ~1.5 years (and it was def very good back then)

Have you played around with other models like flux?

i think v7 is coming out soon. i think it doesn't blow my mind more than it did a year ago really but it's gotten incrementally better but i think maybe they've focused more on improving the website ux cuz that is awesome now