Post #3864709
2026-07-16 16:14 UTC
@brib@bribstodon.xyz that *how* she prompts can steer toward an answer. She went from prompts like "why does x happen" or "is fluoride scary" to "tell me if there are two different arguments for x and show me where it comes from" which tends to get her links, which has pulled her out of some conspiratorial thinking. I believe you can get people out of the belief that these are "answer machines"; I have done so with a lot of not terribly "tech" comfortable people.
Replies (1)
-
@grimalkina@mastodon.social 2026-07-16 16:15
@brib@bribstodon.xyz I do think there is power in a language model, that many people on this platform won't acknowledge. There is a lot of text-based information in scientific literatures that is really difficult for ordinary people to search. There are ways of thinking about infrastructure where LLMs are a part of the pipeline, I think; healthcare systems are fully doing it. A majority of US doctors now use tools like OpenEvidence every single week. So they are present in health regardless of patients