I didn't realize that the Cult of Amma Gyahway was a precursor of the Adeptes Mechanicus. The Omnissiah protects (40k nerd reference if you aren't familiar). Seriously, pretty good for thought provoking fiction and I like the approach. Better than a lot of the 'New war - fan fiction' I've seen (some bad stuff in the Marine Gazette, Mick Ryan). Funny note: I've been on a kick lately about terrible AI pics used as illustration all over Substack. Part of that has been collecting examples for a project. Your pic at the beginning almost got me. Then I thought "Davis is pretty deliberate - this has to be intentional and part of the story".
One last thought - you are on to something. Articles, comments, social media itself is awash with blind reposting of AI generated information. That's an expression of faith in what the machine tells us. There is a clear difference between conditional/critical consideration of the output versus the words of the #mostholyProphetAI. I am surprised we haven't seen GodGPT yet. I know various mILsTAFFAI projects are running.
P.S. If you aren't familiar with the creation of the Medallion Fund (Jim Simons and Renaissance Technologies), you may find the case interesting. It's a precursor to what we are seeing today with AI globally.
The illuminated images are a callback to The Canticle for Liebowitz, Walter Miller's scifi book that was the spark for the structure of the book. If you haven't read it, I highly recommend the book, though it does not have anything to do with AI.
As to the use of AI images, I don't paint with too broad a brush when I describe ML art. There is certainly a lot of low effort single prompt Grok spam being used. But there's also ways to leverage LLMs to generate novel and deliberate images . In the case of these I tried out four different base models and about 20 LoRAs before finding a good mix. I also dug up source images to leverage img-to-img and ControlNet to shape the output. Things like the GeForce 580 that Krizhevsky and his team used for their original research, which is what you see on the left in the third post.
I didn't realize that the Cult of Amma Gyahway was a precursor of the Adeptes Mechanicus. The Omnissiah protects (40k nerd reference if you aren't familiar). Seriously, pretty good for thought provoking fiction and I like the approach. Better than a lot of the 'New war - fan fiction' I've seen (some bad stuff in the Marine Gazette, Mick Ryan). Funny note: I've been on a kick lately about terrible AI pics used as illustration all over Substack. Part of that has been collecting examples for a project. Your pic at the beginning almost got me. Then I thought "Davis is pretty deliberate - this has to be intentional and part of the story".
One last thought - you are on to something. Articles, comments, social media itself is awash with blind reposting of AI generated information. That's an expression of faith in what the machine tells us. There is a clear difference between conditional/critical consideration of the output versus the words of the #mostholyProphetAI. I am surprised we haven't seen GodGPT yet. I know various mILsTAFFAI projects are running.
P.S. If you aren't familiar with the creation of the Medallion Fund (Jim Simons and Renaissance Technologies), you may find the case interesting. It's a precursor to what we are seeing today with AI globally.
The illuminated images are a callback to The Canticle for Liebowitz, Walter Miller's scifi book that was the spark for the structure of the book. If you haven't read it, I highly recommend the book, though it does not have anything to do with AI.
As to the use of AI images, I don't paint with too broad a brush when I describe ML art. There is certainly a lot of low effort single prompt Grok spam being used. But there's also ways to leverage LLMs to generate novel and deliberate images . In the case of these I tried out four different base models and about 20 LoRAs before finding a good mix. I also dug up source images to leverage img-to-img and ControlNet to shape the output. Things like the GeForce 580 that Krizhevsky and his team used for their original research, which is what you see on the left in the third post.