If you know what a ice sculpture of a human looks like and you know what an ice sculputre of a swan looks like you can maybe infer what an “ice sculpture of a flying crocodile with wings” might look like too, by combining what you already know. It’s as much incorrect as your imagination, no?
Just like it can Hallucinate text, it also hallucinates content. That’s a core part of the generative feature
It hallucinates from incorrectly putting other info in its network together. It is all just stochastics.
That is not original or new its is the core of what slop is.
The problem is that it does not have a goal or even just understands why it is doing what it is doing.
If you know what a ice sculpture of a human looks like and you know what an ice sculputre of a swan looks like you can maybe infer what an “ice sculpture of a flying crocodile with wings” might look like too, by combining what you already know. It’s as much incorrect as your imagination, no?