2026-09-24 09:16 UTC
@gavin57@toot.wales yes, apparently was on the hunt for health data
Replies (1)
-
@ApostateEnglishman@mastodon.world 2026-09-24 09:35
@glynmoody@mastodon.social @gavin57@toot.wales You know already, but this will end up being another case of human researchers using the LLM to generate ideas, and the bot ingesting their "private" convos into its training data. This is how it appears to "solve problems", when in reality it's incapable of inferring/inducing anything a human hasn't already published or shared with it. That's just not how they work. Where its dataset is sparse or lacks samples that match the prompt, it confabulates nonsense ("hallucinates").