How To Preserve the Online Information Ecosystem in the Presence of Google AI Overviews

In this article for Pro Market, authors Saharsh Agarwal and Ananya Sen discuss how AI-generated search summaries are reshaping the relationship between search platforms and online publishers. Their field experiment found that the appearance of Google AI Overviews reduced users’ organic clicks to third-party websites by 39.8 percent and increased searches resulting in no clicks by 34.5 percent, without measurable improvements in users’ perceived search quality or ease of finding information.

The authors argue that these findings raise significant economic and policy concerns because AI-generated answers can substitute for visits to the external sources that sustain the online information ecosystem. Although citations within AI Overviews provide some referral traffic, they accounted for only about 8 percent of clicks in the study. As a result, allowing publishers to opt out of AI Overviews may not resolve the underlying problem of lost traffic, while publishers could also face uncertainty about how opting out affects their organic search rankings.

The article considers several implications for competition and copyright law. Because Google controls both the search platform and the placement of its own AI-generated answers, the authors suggest that AI Overviews may raise self-preferencing concerns, particularly when they occupy the most prominent position in search results. The findings may also be relevant to copyright debates because AI-generated answers rely on publisher-produced content while potentially substituting for the original sources, although the authors emphasize that their research does not resolve questions of infringement or fair use.

Finally, Agarwal and Sen argue that effective policy should address the broader economic effects of AI-mediated search rather than focusing solely on publisher consent or compensation. Potential remedies include making source links more prominent, redesigning interfaces to preserve meaningful pathways to original content, and developing licensing or compensation mechanisms. However, they caution that licensing may disadvantage smaller publishers and increase dependence on dominant platforms. Ultimately, policymakers should ensure that advances in AI-powered information access do not undermine the economic incentives needed to produce the high-quality information on which those systems depend…