Guardian Investigation Reveals Misleading Health Information in Google AI Overviews
A Guardian investigation has found inaccuracies in Google's AI Overviews, which provide health information summaries at the top of search results. These AI-generated summaries have been found to contain misleading information, posing potential risks to users seeking health advice.
In one case involving pancreatic cancer, the AI advised avoiding high-fat foods, a recommendation that experts have described as dangerous and potentially harmful. For liver function tests, the AI presented raw numbers with little context, which could lead to misinterpretation for patients with liver disease. Additionally, a search relating to vaginal cancer incorrectly listed the Pap test as a test for vaginal cancer.
The investigation also uncovered inconsistency in AI responses; the same queries produced different AI Overview answers when repeated, as the AI drew from varied sources.
In response, Google stated that most AI Overviews are factual and helpful. The company emphasized its ongoing investment in improving the quality of these features and noted that the accuracy rate is comparable to other search features. Google also assured that they will take action to address misinterpretations.
Health organizations including Pancreatic Cancer UK, the British Liver Trust, Eve Appeal, and Mind have expressed concerns about the potential harm and health risks posed by the misleading AI summaries.
This issue highlights broader worries about the reliability of AI-generated summaries and the confusion consumers may face when accessing health information online.
