It seems like a lot of effort will be spent to ensure that content is not generated by an AI system. This seems very tricky to enforce - a stopgap measure could be to require any person to upload a video where they say their comment before being able to post it to a website. Of course, along with the regression of user experience, such a thing could be tricked by reading an AI generated text, and eventually with deepfakes.

Is it viable to constrain conversation to human voices this way or is the cat out of the bag? Is it desirable? Should we just go back to in-person conversations?