It sounds the big problem is that AI systems could very easily pretend to be human, with no clear way to know exactly who or what you’re talking to. Video is currently much harder for AI to render than text – would increased video calling be at least a stopgap solution to know you’re talking to a person?
.
