Theoretically it could work but I guess depending on the prompt you would introduce bias in the vast majority of cases. > ensuring [ethical] considerations I

Theoretically it could work but I guess depending on the prompt you would introduce bias in the vast majority of cases. > ensuring [ethical] considerations I guess we differentiate synthetic/fraudulent data depending on the qualifier like "ethical". Without bias, they are the same. It will get interesting if "synthetic" data in models will be used for analysis. I doubt these processes will be too ethical.