[mistermann]

I've been doing some experimentation in ChatGPT with this, and it is amazing.

With just slightly different ChatGPT query prompts, extremely different, conflicting narratives for a given event can be produced, and each of them sounds/seems perfectly reasonable, and True.

For someone to be invulnerable to this sort of thing, they would have to have fairly substantial background in several different obscure domains (none of which are taught in school, quite conveniently), as well as violate several strongly enforced social norms (ie: engaging in "pedantry", thinking in ways we've been trained to pattern match to conspiratorial thinking, falling for Russian propaganda, etc).

It does not surprise me in the least how much polarization there is in society over what is actually going on, and how each side genuinely believes that they are take on it is correct. It seems (to me) almost like the system has been deliberately designed to produce this output, because it certainly isn't that hard to figure out how to vastly improve on it.