Fun experience and also some really great reading at the end that comes from A podcast i highly recommendTHE PROMPT: I gave it a 7200 token raw input transcription. I asked for markdown formatting, good section titles, bolding important words, good quote formatting, write every word as is and to correct obvious spelling mistakes (could be seen as conflicting rules) also to give a recap of spelling mistakes fixed.

If you're curious on the methodology Get raw information about sources that document key aspects of life. (Check other personal sources... of which there were none)Get raw information about sources that document key aspects of life. (Check other personal sources... of which there were none)Transcribe and internalize the 45 sources documenting his life in records (22k tokens turned into a 9k token document)Transcribe and internalize the 45 sources documenting his life in records (22k tokens turned into a 9k token document)use the documentation to do a deep research and focus on history of locations and what happened with his familyuse the documentation to do a deep research and focus on history of locations and what happened with his family

Mary Jane Bargery (1862–1948)This document represents hours of work, use of research by lots of family members and historians, transcriptions by unamed masses and some really well placed ai prompts to help get it all organized. Staring an o3 Deep Research prompt uncovering a lot of interesting insights and new resources

IT MAKES ME HAPPY I love creating, I love art, I love trying to be good at new things, and I adore Jesus. I also appreciate the new tools provided by ai and so of course I'm making music albums and putting them on spotify and paintings and putting them on my website. So... when added ai image creation last week and it is much higher quality than midjourney which I've used for a long time now. I of course am gonna test with creating Jesus art.

I may decide to stop doing all 9 models in these polls when i request this much text. I have done enough requests to know it's usually Recraft, Ideogram, Pro then Ultra then Dev ... that's the order of accuracy. So this may be one of the last where i use all 9 models when i have this much text. BUT running these blind tests are still great to do so it's not just matter of theory but becomes tested.

THE PROMPTAn illustration to be a blog post thumbnail cover that represents a blind vote on ai art. Text that says "Blind Vote for best Image Model"EDIT: Didn't release this as a poll type post so you'll need to vote here: