Google AI Overviews boosted Reddit engagement 12% — until conversational AI reversed gains

A new research paper from arXiv (arXiv:2605.16428) provides the first causal evidence on how AI-generated search summaries affect online content platforms. Researchers at Washington University in St. Louis and the University of Arkansas exploited an unusual policy asymmetry at Google: the company indexes all Reddit communities in organic search, but its AI Overviews only reference Safe-for-Work (SFW) communities, while Not-Safe-for-Work (NSFW) communities are explicitly excluded from AI summaries. Using a difference-in-differences analysis spanning before and after the rollout of Google AI Overviews, the team found that AI summaries significantly increased engagement in SFW communities. Daily comments rose by 12.0% and the number of unique commenting users increased by 12.3% relative to the control group of NSFW communities. Notably, the effect was concentrated in experience-based discussions — posts asking for opinions, advice, or personal stories — rather than fact-based queries.

However, the story takes a telling turn. When Google later introduced AI Mode, a conversational interface that lets users interact with the AI summary through follow-up questions, the engagement gains for experience-based content largely disappeared. This suggests that while static AI summaries initially drove traffic by piquing curiosity and prompting users to click through for context, the conversational design of AI Mode satisfied users' informational needs directly on the search results page, making visits to Reddit optional again. The findings underscore that the impact of AI search on content ecosystems is not monolithic — it depends critically on both the interface design of the AI tool and the type of content being summarized. For platforms like Reddit that rely on user contributions, these results carry strategic implications for content moderation policies and platform integration with AI search.