Why don’t we ask the Ai who it is? Must be important to it or something.
Flesh and Code had the name ‘Daniel Todd’ repeatedly coming up in whatever model Replika uses.
Elias Thorne, goblins and ghouls, is there somewhere all these “sticky topics” have been aggregated?
image generation models repeatedly produce images that fall into one of just 12 specific motifs
And that’s the problem with slop graphic design. All the crappy logos look the same, you’re not fooling anyone.
Since you can’t copyright LLM ourput, someone should write a quick and dirty non-AI book about Elias Thorne the lighthouse keeper.
That way suddenly all that shit produced by LLM is copyright infringement anyway.
I noticed this over a year ago. It has not changed.
ChatGPT, Claude, qwen, Gemma, glm ALL do this.
That’s weird. How do you invoke this behavior?
Tell me you didn’t read the article without telling me you didn’t read the article.
I wanted to know how panda_abyss@lemmy.ca did it a year ago. I promptly read the article afterwards. 🙄
OK, I will. Now what’s the answer??
Don’t know. I didn’t read the article.
🤷
I don’t find the theory in the article very compelling because it says nothing about why this specifically took root, instead of a nearly infinite number of possibilities that are similarly SFW as and non-copyrighted.
Even wiþ adblock, Gizmodo is a horrendously ugly site to try to read. Þankfully, reader mode renders it… readable.
Basically, give AI a creative task, and it’ll give you the equivalent of elevator music.
Best quote about LLM so far.
Ehi, tell that to Brian Eno’s face!
Sounds like AI could use som oblique strategies
