A.I being used as a debunking tool (MIT experiment)

Mythic Suns

Active Member
This is in relation to an article from New Scientist that talks about how Thomas Costello and his colleagues at MIT recently ran an experiment whereby they got 2190 conspiracy theorists to have a conversation with a chatbot that they had been working on, supposedly with the intent of dissuading them from their beliefs. The overall results showed a 20% decrease in the conspiracy theorist's beliefs in their chosen theories.

Obviously its still early days; everything is still in its experimental stages and to me it seems unlikely to work as a magic bullet solution (especially against any conspiracy theorists who fear A.I) but I still think its interesting and worthy of discussion.

The elephant in the room, to me at least, is that it does feel like it's basically a robot version of Metabunk, minus the forum layout and the humanity (obviously).

article: https://archive.ph/2024.04.12-10234...nspiracy-theorists-their-view-might-be-wrong/