Post #3864711
2026-07-16 16:10 UTC
@brib@bribstodon.xyz ...measurably less conspiratorial than the websites and forums she was on before. That has actually given her interesting counterpoints to a lot of the arguments she encounters. I certainly don't think that outcome is inevitable, because I think you could also use AI to convince yourself *of* conspiracy. I find work like this really interesting, because we have always used techniques like journaling or static brochures for persuasion (https://pubmed.ncbi.nlm.nih.gov/39264999/).
Replies (1)
-
@grimalkina@mastodon.social 2026-07-16 16:11
@brib@bribstodon.xyz I think this effect probably goes both directions. It's powerfully persuasive in one direction when it's grounded against clinical and scientific consensus; I can easily imagine it being persuasive in the other direction. That is what the psych intervention research is finding so far. Now, what might help? I've spent some time trying to provide that to the patients I know and talk to. For example, a mental model of an LLM as a text & prediction tool helps them, my friend now sees