AI assistants don’t treat websites equally. They pick favorites, and those favorites often don’t match what people actually search for.
Most people think AI assistants pull answers from the same websites we trust in regular search. But they don’t. This article uncovers which sites are shown more or less often by Google AI Overviews, ChatGPT, and Perplexity-revealing bias that could affect who gets seen and who doesn’t.
The article looks at how three major AI tools-Google AI Overviews, ChatGPT, and Perplexity-choose which websites to show in their answers. Instead of only counting how often a site is mentioned, the study compares that number with how often the site should appear based on how popular its topics are. This shows if AI tools are favoring or ignoring certain websites.
The findings show that Google’s AI mentions Reddit, Quora, and YouTube far more than expected, while sites like Wikipedia and trusted health sources appear less often. ChatGPT also under-uses Wikipedia and Reuters. Perplexity, on the other hand, is the most balanced, showing websites close to how popular they are in regular search.
The main idea is that search is changing fast. It’s no longer just about ranking in Google. If websites want to be seen, they may need to think about how AI assistants choose what to show-and start adapting to that new world.