Same test on 400 pairs: judges picked the wordier site 66% of the time with real words, 64% with x's, and 56% with the text hidden, close to people (54%). If you grade design with an AI judge, it'll push you toward busier screens than people want.
AI judges favor busier designs than people do in 400-pair test
AIDesign Arena ran the same test on 400 site pairs and found AI judges picked the wordier site 66% of the time with real words, 64% with placeholder x's, and 56% with text hidden. Human raters chose the wordier site 54% of the time.
Source details
DesignArena · x.comPublished · added here
Original title: Same test on 400 pairs: judges picked the wordier site 66% of the time with real words, 64% with x's, and 56% with the text hidden, close...