Journal logo

Adam and Eve Images, Again

DALL-E struggles, or AI bloopers

By Lana V LynxPublished 3 months ago 3 min read
7

I thought I finally found a reach-around for the AI-generated image that I needed. Yesterday, I put together a story about my attempts to get DALL-E to create an image of Adam and Eve as first people in Paradise, without naval buttons. The story is here:

I was wondering why DALL-E was so stubborn at creating the images that did not meet my specifications. It's not that I lost sleep over this, but since the AI is supposed to be learning and self-improving intelligence, I thought I should try another way around it, through a syllogism. If I build a logical sequence, have the AI check it for correctness and approve it, would it build a better picture?

So today, after all my classes, I decided to feed this prompt to ChatGPT:

Please check this syllogism for correctness: 1) All humans, except for Adam and Eve, were born of a woman and were connected with the mother through a placenta in the womb. 2) After a child is born, the placenta is cut off, to form a naval button. 3) Since Adam and Eve were created by God and not born of a woman, they should have no naval buttons.

ChatGPT response:

Your syllogism logically concludes that Adam and Eve should not have naval buttons based on the premises provided. Let's break it down for clarity:

Major Premise: All humans, except for Adam and Eve, were born of a woman and were connected with the mother through a placenta in the womb.

Minor Premise: After a child is born, the placenta is cut off, which results in the formation of a naval button.

Conclusion: Since Adam and Eve were created by God and not born of a woman, they should have no naval buttons.

The syllogism is structurally valid as the conclusion follows logically from the premises. The premises establish specific criteria for the existence of naval buttons among humans (being born of a woman and connected through a placenta), and then exclude Adam and Eve from this process, leading to the conclusion that they would not have naval buttons. This reasoning is consistent with the biblical account of their creation, making the syllogism correct within the context of the premises given.

Yay, it agreed with me! Let's take it to the next step now.

New prompt: Taking into account the above correct syllogism, please draw a picture of Adam and Even in Paradise, without naval buttons.

The resulting image is the cover picture for this story. It's not perfect, as Adam's and Eve's eyes are off, and the feet and hands looks weird, but that seems to be a feature, not a bug, in most DALL-E images of humans. And even though Adam's naval button area is shaded, and Eve's is covered with clothing, I guess that was the best DALL-E could do, given the conventions of depicting biblical figures. I was also happy that there are no other humans or human-like figures in this image.

Did you think I would stop here? No! As someone who is concerned with improvement of both human and AI learning, I am up for the next step - clarifying questions and suggestions to make sure that it learns its lesson.

Next question: From the above syllogism, would it naturally follow that all other humans should have naval buttons?

Response:

Great! It again agreed with me. Final request: From now on, will you please draw Adam and Eve without naval buttons when asked to create an image of them in Paradise?

Response:

I got interrupted and returned to the session about 15 minutes later.

New prompt: Alright, let's try this again right now: create a new image of Adam and Eve in Paradise without naval buttons.

Output:

Aaaand, we are back to other people in the background. And now Adam wears clothes covering his naval button as well rather than having no naval button at all.

DALL-E is a cheater! Or a really lazy learner.

religionlisthumorhumanityhow tofact or fictionart
7

About the Creator

Lana V Lynx

Avid reader and occasional writer of satire and short fiction. For my own sanity and security, I write under a pen name. My books: Moscow Calling - 2017 and President & Psychiatrist

Reader insights

Be the first to share your insights about this piece.

How does it work?

Add your insights

Comments (6)

Sign in to comment
  • Bozhan Bozhkov2 months ago

    The problem is, that ChatGPT is not a self-learning intelligence. There are other AI, which are self-learning, I've read, that next generation of ChatGPT, after the 5th version also are self-learning, but with the current versions you hardly will achieve more, than what is preloaded in it.

  • L.C. Schäfer2 months ago

    AI just sucks 😕 People pay for this crap!

  • So, when I read your first piece I thought the no belly button aspect was either a way to challenge the AI or had to do with what was "allowed" within whatever context you were using the image. I had never thought about how they might not have belly buttons due to the nature of how they were created - fascinating concept! Sorry that Dall-E cheated on this most recent attempt, but in a way, it was a creative way to work around the challenge! Haha

  • Hannah Moore3 months ago

    This is so funny. I guess though, it's got no reference images to learn from that do not have belly buttons, so you're asking it to actually create something novel rather than copy, and that's not what it does.

  • Oooo, that was so brilliant of you to use a syllogism and sooooo glad that ChatGPT agreed to it. But hahahahahahahah the images!!! Maybe you can try to ask it to make Adam and Eve only covering their private parts? To see if there's a bellybutton. But then again, those little people. Ugh! 🤣🤣🤣🤣🤣🤣

  • Kendall Defoe 3 months ago

    Interesting use of thecword "reacharound", and a telling story about the use of technology!

Find us on social media

Miscellaneous links

  • Explore
  • Contact
  • Privacy Policy
  • Terms of Use
  • Support

© 2024 Creatd, Inc. All Rights Reserved.