The latest version of the artificial intelligence (AI) image generator Midjourney can finally represent human hands with a full set of five fingers. Version 5 of Midjourney was released yesterday, March 16, and while the update isn’t always perfect, it marks a definite improvement for emerging technology. It also eliminates one of the last telltale signs of an AI-created image.
For all their mind-blowing abilities, AI image generators just couldn’t figure out hands. Items appeared explaining the phenomenon and education users how to fix it. Other people online made jokes about the issue (“Generative AI not being able to draw hands is the most relevant thing it has going for it,” @MNateShyamalan wrote on Twitter). While version 5 still produces a fair number of anatomically impossible arms and fingers, the update has rendered hand talk largely obsolete.
A virality Tweeter yesterday, March 16, took a more sinister approach. “Just a warning – Midjourney’s AI can now do hands correctly. Be very critical of any political imagery (especially photography) you see online that tries to elicit a reaction,” wrote @TheCartelDel, a self-proclaimed 3D character specialist.
Many viewers have already proven unable to tell the AI-generated artwork from the real deal. A few weeks ago, a popular Instagram account revealed his “photographs” – said to have been taken with a Nikon D810 – were actually created with AI prompts. (These portraits, in particular, did not represent hands.)
In response to @TheCartelDel’s tweet, a commentator pointed out that the new five-picture version of Midjourney is still “in the weird valley.” Hyperallergic encountered a few troubling traits when testing the new update, such as the appearance of rubbery skin, but overall the results looked much more realistic than their version four counterparts.
The fear of deceptive misuse of AI-generated images is just the latest controversy surrounding the technology. The artists pushed back against programs such as Midjourney, citing that they “scrape” the internet for existing images to create new works. Creators have cited plagiarism issues, and big companies have also fought the new technology. Getty Images, for example, is pursue Stable Diffusion (another popular AI image generator) alleging intellectual property rights infringement.
