Artificial intelligence had a big, bizarre weekend. A Google engineer told the Washington Post he believes one of the company’s AI chatbots is sentient, and Twitter users discovered DALL-E, “an AI model that generates images from any prompt you give.” While I cannot entertain the possibility of the former for fear that my head will explode, I can (and did) entertain myself with the latter for hours!
For the uninitiated, DALL-E was trained to convert text into images using “unfiltered data from the Internet” — meaning you can input any word or phrase you like, and the DALL-E generator will virtually visualize it for you. (Ex: “nosferatu in rupaul’s drag race”; “james corden mugshot”.) I simply had to see what the program would make of beauty product propaganda, so I pulled a bunch of phrases from marketing emails I received this weekend and ran them through DALL-E.
Here’s how AI interprets PR BS:
1. “a much needed boost of hydration to the skin”
2. “a multi-benefit skincare treatment that celebrities like Gwyneth Paltrow swear by”
3. “pro-aging philosophy”
4. “constricts muscles so much that the wrinkles disappear”
5. “a second skin jell”
6. “like a drink for your cheeks”
7. “The Fire and Ice Facial”
8. “shrink the look of pores”
9. “the Savior to Your Skin”
10. “body cosmetic procedures are both a science and an art”
11. “exciting new vegan Collagen lip launches”
12. “Secure Attachment Comfort Serum sheds light on the psychological theory of attachment”
(Like… can you believe this is in reference to SERUM??!? AI can’t!)
The art I didn't know I needed
It’s odd (creepy) how Gwenyth Paltrow looks altered in each pic but somehow exactly the same. Not convinced she’s truly human