It’s not just a matter of discomfort for something new, but at something highly dangerous. Deepfakes have several bad and disturbing use cases, like itentity theft, sexual exploitation, marketing abuse, political manipulation, etc. In fact, I hard to find a significant good use of such technology.
It would be nice if we could choose our bias. Sometimes, we might want it biased towards scientific sources, sometimes, towards user-generated content, sometimes towards institutional sites, etc.