Nice Try, ChatGPT: Accusations of AI authorship are increasingly weaponized in online political discourses, selectively dismissing opposing arguments and severely disrupting debates. It is becoming increasingly common to encounter AI-generated text online. Yet at the same time it is often difficult to distinguish authentic text from generative outputs, particularly within shorter posts. Given this conflict, we expect accusations that a text was AI-generated to become a common strategy for deflecting unwelcome arguments. To fill this gap we draw on a complete dataset containing two years’ worth of public archives from across twelve political Reddit communities, enabling us to identify an estimated 8,000 accusatory comments involving either bot or AI-generated text accusations. Our results suggest that AI accusations are increasingly being deployed online to disrupt political debates and that this most often serves as a motivated attempt to dismiss counterattitudinal arguments. While these accusations did not drive their targets to withdraw, they were uniquely disruptive, with substantive engagement following these accusations falling to less than half of what we found after more widely discussed hostile dismissals.. As a result, in many cases the oppositional arguments presented by accused accounts were left entirely unanswered. These results provide behavioral evidence of what we describe as the motivated attribution of inauthenticity, where accusations are used to selectively dismiss opposing views. These findings both build on and depart from the recent study of bot accusations. Bot” has shed its technical meaning and become a tool of dehumanization. While we found bot accusations to be more disruptive than other types of dismissals, AI accusations diverted conversation at nearly twice their rate. Our findings show that the mere knowledge of generative AI tools can generate similar forms of epistemic uncertainty in everyday political conversations even when it is not actually being used. As AI-generated text becomes more common, this work suggests that debate may become more challenging as disagreements can be diverted toward debates about who, or what, wrote an oppositional argument, just as warnings about deepfakes have been shown to influence the dismissal of genuine videos.

Only 26% of men think college is “very important” compared to 40% of women (down 19 & 17 ppts respectively) news.gallup.com/poll/715133/… #