KI-Chatbots können die Einstellung der Wähler zu Kandidaten und Richtlinien verändern. Die KI-Modelle versuchten tendenziell mit Fakten und Beweisen zu überzeugen. Allerdings waren nicht alle KI-Modelle genau. Die KI-Modelle, die sich für Kandidaten der politischen Rechten einsetzen, machten ungenauere Behauptungen.
https://www.nature.com/articles/s41586-025-09771-9
3 Kommentare
>We observed significant treatment effects on candidate preference that are larger than typically observed from traditional video advertisements, We also document large persuasion effects on Massachusetts residents’ support for a ballot measure legalizing psychedelics. Examining the persuasion strategies[^(9)](https://www.nature.com/articles/s41586-025-09771-9#ref-CR9) used by the models indicates that they persuade with relevant facts and evidence, rather than using sophisticated psychological persuasion techniques
seems a bit weird to compare this to a video ad. With a chatbot you’re having a full conversation, wouldn’t it be better to compare it to an equivalent conversation with a human?
I’ve never understood this, for me nothing anyone says about a political party can change my mind, i see what their actions are and if they just talk or they actually try to do stuff, and then i vote for them, an AI telling me hey look they are bad wont make me change my mind.
It doesn’t surprise me that a piece of software trained on data made by humans would skew the same way the humans do.