Google’s AI Health Overview Faces Scrutiny
An investigation found that Google’s “AI Overview,” displayed prominently in search results, was misleading users by delivering inaccurate and potentially harmful medical information. This included incorrect advice on liver function tests. Following the discovery, the search engine decided to remove the AI feature from their platform.
Google announced the removal of its AI-generated health diagnostics tools after it became evident that they were disseminating incorrect information to users. The investigation highlighted serious worries regarding the reliability of health content produced by the “AI Overview.”
The AI Overview tool employs generative artificial intelligence to provide snapshots of information based on user queries. These summaries appear at the top of search results and were previously touted by Google as a dependable source for quick responses.
However, the investigation revealed significant inaccuracies in some health-related summaries, which could pose risks to users. A particular case involved misleading information concerning liver function test results, leading experts to label it as both dangerous and alarming due to its implications for patients.
When searching for normal ranges for liver blood tests, users encountered numerous figures paired with little context. It failed to consider critical factors such as patient demographics—nationality, gender, ethnicity, and age—which all influence test outcomes.
Medical professionals pointed out that what Google’s AI considers “normal” can vastly differ from actual medical standards. This misalignment presents a serious risk that individuals with severe liver conditions might misinterpret abnormal results as normal, potentially skipping necessary follow-up appointments.
In response to the investigation findings, Google removed AI-generated summaries linked to searches asking about normal liver test ranges. While a spokesperson noted they typically don’t comment on individual content removals, it was emphasized that where AI summaries lack context, the company intends to make broader improvements.
Sue Farrington, president of the Patient Information Forum, commended the decision to remove these summaries while acknowledging ongoing concerns about the reliability of AI-generated health information. She viewed this action as a positive step, yet highlighted the need for continued efforts to enhance trust in Google’s health search results.
Farrington remarked that millions globally struggle to find trustworthy health information, stressing the importance of Google directing users to reliable, evidence-based medical resources.
The investigation also discovered other AI-generated summaries still active on the platform that experts deemed erroneous and potentially dangerous, particularly those discussing cancer and mental health. Despite the issues identified, these summaries are set to remain in search results.
When asked about the failure to remove these problematic summaries, Google advised that it links to reputable sources and alerts users when they should seek professional guidance. A spokesperson mentioned that an in-house team of clinicians reviewed the information and concluded that much of it was accurate and supported by high-quality resources.
