Concerns Over Google AI Health Summaries
A recent investigation by The Guardian has raised alarms about the potential dangers posed by false health information in Google’s AI-generated summaries. These AI Overviews are designed to offer quick insights on various topics, including health-related queries, with Google stating they are “helpful” and “reliable.”
However, some summaries, prominently displayed at the top of search results, have delivered misleading and inaccurate health information, putting individuals at serious risk. For instance, in a concerning case, Google recommended that people with pancreatic cancer avoid high-fat foods. Experts noted this advice could be detrimental, actually opposing what is generally recommended and possibly increasing mortality risks.
Another troubling instance involved incorrect information regarding crucial liver function tests, potentially leading individuals with serious liver conditions to mistakenly believe they are in good health.
Further misguidance was evident in searches related to women’s cancer tests, with experts commenting that the inaccuracies could result in individuals overlooking significant health symptoms.
A spokesperson for Google remarked that many instances cited were “incomplete screenshots,” but from their assessment, they linked to “well-known, reputable sources” and emphasized the importance of seeking expert guidance.
This investigation comes as concerns grow about AI confusion, particularly as consumers may mistakenly view AI-generated content as trustworthy. A study from last November indicated that AI chatbots often provided inaccurate financial advice, highlighting similar worries about summaries of news stories.
Sophie Randall, director of the Patient Information Forum, expressed worry that Google’s AI Overviews could inadvertently place inaccurate health content at the forefront of online searches, compromising public health safety. Similarly, Stephanie Parker, director of digital at Marie Curie, pointed out that when individuals confront distressing situations, the risk of receiving misleading information could significantly harm their well-being.
The Guardian’s findings stemmed from numerous raised concerns from health professionals and organizations. Anna Jewell, director of support at Pancreatic Cancer UK, called the advice to avoid high-fat foods “completely incorrect,” stressing that following such guidance could limit an individual’s caloric intake, leading to adverse treatment outcomes.
Inquiries regarding blood test ranges for liver function also produced misleading statistics, lacking the necessary context regarding nationality, gender, ethnicity, or age—factors essential for accurate interpretation. Pamela Healy, CEO of the British Liver Trust, highlighted the risks involved, particularly since many liver disease patients show no symptoms until advanced stages, making accurate testing critical.
In another example, a search for “vaginal cancer symptoms and tests” inaccurately included pap tests as a detection method for vaginal cancer. Athena Lamnisos, CEO of the Eve Appeal cancer charity, underscored how misleading information could deter individuals from seeking necessary medical evaluations for potential symptoms.
Google’s AI Overviews have also misrepresented information on mental health conditions. Stephen Buckley, head of information at Mind, noted that some AI summaries provided “dangerous advice” and could dissuade individuals from seeking needed help, often lacking important context or nuance.
In response, Google maintains that most of its AI Overviews deliver factual and useful content, continually striving for quality improvement. They pointed out their accuracy rates align with other long-standing search functionalities while committing to address any misinterpretations or contextual omissions in the summaries.
A Google representative stated, “We invest significantly in the quality of AI Overviews, particularly for health topics, and the vast majority provide accurate information.”
