I’ve been following the developments around Google’s AI health summaries, and recently, I stumbled upon a study that really caught my attention. It turns out that Google’s AI often relies on sources like YouTube, which aren’t always thoroughly vetted by medical professionals.
Most of the health advice summarized by Google’s AI Overviews is sourced from non-medical platforms, raising concerns about the potential for misleading or unsafe information on critical health topics. This revelation has prompted discussions about the reliability of these sources.
What’s happening: According to The Guardian, Google’s AI Overviews occasionally showcase incorrect or even risky health information, a claim supported by medical charities and experts who reviewed this content.
Examples of such cases include incorrect advice about diets for pancreatic cancer and misleading interpretations of liver blood test results. Google, however, stands by its AI, arguing that the instances were taken out of context and that their Overviews generally link to trustworthy sources.
The bigger issue: Our friends at SE Ranking conducted an in-depth analysis of where AI Overviews derive their information. They assessed 50,807 health-related searches in Germany, unveiling a significant insight.
Nearly two-thirds of the citations for Google AI Overviews originate from sources that lack strong medical or evidence-based credibility, highlighting a concerning trend.
By the numbers: YouTube emerges as the top-cited source for these AI-produced health summaries, accounting for a whopping 4.43% of all citations. This percentage is considerably higher than that of more traditional medical sources like hospitals or health associations.
In total, 34.45% of citations come from more reliable medical sources. Interestingly, academic journals and government health entities only make up about 1% of citations. This reliance on video content is notable; despite YouTube’s first-place standing in AI citations, it ranks just 11th in organic search results.
There is also a misalignment in source credibility, as only 36% of pages cited by AI appear in Google’s top 10 organic search results.
Why we care: As someone deeply invested in public safety, I find it crucial to highlight that Google’s AI Overviews serve as a primary layer of health information, especially for sensitive topics like Your Money or Your Life issues. Given that over 82% of health inquiries prompt these AI-generated summaries, ensuring their quality is paramount for public well-being. Google must be held to the same rigorous standards it demands from others in this domain.
The study: If you’re interested in delving deeper, check out this insightful study that compares the frequency of YouTube references in AI Overviews against those from reputable medical sites.
Inspired by this post on Search Engine Land.
