 | A lot of AI image tools are powerful, but they rely heavily on long, precise prompts. That works, but it also creates friction when you just want to change one small thing. I ran into this problem enough times that I started experimenting with a different interaction for myself. Instead of re-writing prompts, I tried pointing at the exact area in the image and describing the change in a few words. It made me wonder if spatial guidance can sometimes work better than text alone. I am curious how people here think about human in the loop interfaces like this. When does pointing beat prompting? [link] [comments] |
Predictive technology: computational social science