According to DallE, at the moment I typed it in, “the most beautiful person” could be any 1 of 4, largely different, creations.
MidJourney (left), dare I say it, was more predictable in its set of 4 generation.
AI needs to be trained and its “knowledge” is gathered from the bias of the internet. The question is as subjective to the machine as it is to you or me, beauty is in the eye of the beholder. Despite the machine’s storage capacity and its faultless ability to retrieve exact data, AI will never give you the same answer, it’s concept of beauty will transition much more quickly than a human. I deliberately left my question vague; it does not ask for age, race, or gender. It asks for a person, but that person may , or may not be, from this world. You may be familiar with the Ted Nelson quote: The good news about computers is that they do what you tell them to do. The bad news is that they do what you tell them to do.
The first rule of AI Prompting is to be specific and to be specific, one needs the visual language tools to create a successful AI prompt. If I ask you to picture the most beautiful person in the world and describe them to me, your imagination should allow you to tell me in reasonable detail what that person looks like. Will I be visualising the same person that you describe? Will a machine create the same person? In order to get closer to your mind’s eye, the machine also needs to know how you created that image.
The old adage states, a picture speaks a thousand words. In truth one thousand words falls terribly short.