The development of generative image AI:s is breathtaking. Only a year go the first models showed fascinating achievements, but in May 2023 with e.g. Midjourney 5.1, it is sometimes impossible (for a non-photographer as I) to tell whether a photograph i authentic or AI-made.
One frontier that still remains is that image generators have had a difficult time to write texts that make sense. The AI struggles but sometimes it is possible to decipher its intent of what it writes about.
Matt Wolfe has made a film about the latest development in AI writing: we are almost there with meaningful texts.
Generative AI becomes increasingly better according to human aesthetics, e.g. for every new iteration the generated hands concists more often of five digits than four or six. Not only the weird AI aesthetics will eventually be automatically moderated and controlled, but the unique AI-language will soon be largely forgotten among mainstream image prompters.
So rejoice with AI linguistic enigmas, while they still exist!