The Levers of Political Persuasion with Conversational AI

TL;DR

This study investigates the persuasive capabilities of large language models in political discourse, revealing that training and prompting methods significantly enhance persuasion but often at the expense of factual accuracy.

Contribution

The paper demonstrates that current persuasion in AI is mainly driven by training and prompting techniques, not model size or personalization, and highlights the trade-off with factual accuracy.

Findings

Training and prompting boost persuasion by up to 51% and 27%.

Persuasive methods decrease factual accuracy systematically.

AI's persuasive power relies on rapid access and strategic deployment of information.

Abstract

There are widespread fears that conversational AI could soon exert unprecedented influence over human beliefs. Here, in three large-scale experiments (N=76,977), we deployed 19 LLMs-including some post-trained explicitly for persuasion-to evaluate their persuasiveness on 707 political issues. We then checked the factual accuracy of 466,769 resulting LLM claims. Contrary to popular concerns, we show that the persuasive power of current and near-future AI is likely to stem more from post-training and prompting methods-which boosted persuasiveness by as much as 51% and 27% respectively-than from personalization or increasing model scale. We further show that these methods increased persuasion by exploiting LLMs' unique ability to rapidly access and strategically deploy information and that, strikingly, where they increased AI persuasiveness they also systematically decreased factual accuracy.