Okay, that was a read. What is the threshold for "AI code?"
- Hey AI agent, write me a "Hello World" in gdscript!
- Agent: `print("Hello World!")`
Oh, whoops, one line of AI generated code ended up in my game. I highly doubt that automatically makes it slop. Before anyone laughs, what if I ask my agent how to call a signal? What if it's not my agent, but an auto-generated answer embedded in my search engine, and now I'm contaminated with the exact information I wanted from a source someone will hate me for being honest about spotting out the corner of my eye (or front and center for half the if you use the G-search giant).
Now, if I were delegating entire systems and shaders for my agent to write, then sure, that would make for a greater chance of slop if I didn't understand it line-by-line.
Oh, get this gripe! As I understand the tags, I could prompt an AI to generate a slop game concept, then abstain from using AI output in my game proper. See, I was doing some research recently, and I saw something about how AI used at the beginning of the creative process produce more AI-like results --even when completed by a human-- than compared to an AI following up on a human's creative session. Yet the more damaging and likely to produce slop is the one we aren't going to stigmatize.
I'm quickly developing the philosophy that "Art is worth the time you put into it." If a knowledgeable artist spends a week on an AI image, massaging ControlNet into submission to produce a premium picture that means something to him and his audience, I would be more likely to value that above the aftermath of an afternoon throwing darts at a canvas with paint balloons tacked on.
Again: what is the threshold for "AI code?"