[eudamoniac]

I think people will eventually learn to not trust any videos or stories they see online. I think the much bigger issue will be what happens when the LLM providers encode "alignment" into the models to insist on certain worldviews, opinions, or even falsehoods. Trust in LLMs and usage of them is increasing.

"Great question! No, we have always been at war with Eurasia. Can I help with anything else?"

[NoMoreNicksLeft]

"Eventually" does alot of heavy lifting in your prediction. This is like saying that if you feed poison to panda bears, they will eventually become immune to poison. On what timescale though? 8 million years from now, if the species survived, and if I've been feeding that poison to each and every generation... sure.

If I just feed it to 10 pandas, today, they're all dead.

And I suspect that humanity's position in this analogy is far closer to the latter than the former.

[eudamoniac]

People stopped falling for photoshopped pictures and staged Chinese reels pretty quickly. I think people will pretty quickly decide anything outrageous is probably AI. And by people I mean the right half of the bell curve, which is all you can hope for. The left half will have problems in the world as they always have.