Easy to Produce, Hard to Persuade: The Asymmetric Effects of AI on the Online Information Ecosystem

05 May 2026, Version 1
This content is an early or alternative research output and has not been peer-reviewed at the time of posting.

Abstract

The hype around large language models and other forms of AI has led to widespread concern about their effects on information exposure and persuasion, but these fears are likely to outstrip reality. AI is more likely to reinforce existing patterns of exposure and behavior than it is to transform how people understand and relate to the political world. Even with widespread AI use, we expect that most people will continue to consume relatively little political news and that it will be difficult to durably change public opinion at scale. The effects of AI on politics are likely to be greatest in the way it relaxes constraints on monitoring and production for lower-capacity actors and in how state control over AI development risks embedding regime-preferred narratives into the technology. We conclude by discussing how AI enables new approaches to studying information exposure and behavior.

Keywords

AI
artificial intelligence
large language models
LLM
generative AI
public opinion
persuasion

Comments

Comments are not moderated before they are posted, but they can be removed by the site moderators if they are found to be in contravention of our Commenting Policy [opens in a new tab] - please read this policy before you post. Comments should be used for scholarly discussion of the content in question. You can find more information about how to use the commenting feature here [opens in a new tab] .
This site is protected by reCAPTCHA and the Google Privacy Policy [opens in a new tab] and Terms of Service [opens in a new tab] apply.