Hey, I know this is quite delayed but I can clear some stuff up. TLDR you’re both right but we still aren’t able to make it work without compromising on things we aren’t prepared to.
The short version is that there are 2 main ways of setting a cursor image in unity. The “good” way and the “bad” way. The good way is more efficient and sticks to the cursor location perfectly. The bad way can lag slightly behind, this gets worse in the event if a lag spike.
The good way does not allow for a set of images to be used and switched between. It only gives the option for one image. This image is baked at build time and does require rebuilding the entire game to be changed.
The bad way allows for any number of images which can be switched out on the fly. I believe that this method is what chatGPT is referencing. So while it is not technically wrong, it’s not understanding the situation as a whole.
We are still attempting to find a way to use multiple image options in the good way as we are not prepared to take the plunge into the bad way.