Google removes AI health summaries after safety concerns

Google is removing some of its AI-generated health summaries after an investigation finds that users’ health may be at risk by false and misleading information generated using such tools. The feature that was raised about Google AI Overviews appears at the top of search results and provides snapshot answers using generative AI.

This decision came after searches concerning liver blood test results threw up misleading and inaccurate data. Experts caution that the resulting summary from the AI technology may mislead patients with severe liver conditions by giving them a false sense of normalcy.

In other situations, individuals might think that their test results are normal and ignore post-result follow-up care. Google AI Overviews, healthcare misinformation, and online safety are at the focal point of this issue.

As reported in the investigation, “searches such as ‘normal range for liver function tests’ show long lists of numbers in a way that is not properly put into context,” said Hebditch of the British Liver Trust, who declared the information “dangerous” because of the severe consequences of misinterpretation of the information given.

Google has since removed AI Overviews on some liver-related search queries. Nevertheless, a search using a similar question but a different set of words still allows access to an AI summary. According to Hebditch, it raises a great concern, particularly since liver function tests are complex and not so simply interpretable.

Patient Information Forum Chair Sue Farrington welcomed the removals but said trust in health search results is still fragile. She emphasised how millions can't currently access reliable health information and that Google needs to prioritise evidence-based sources.

Google said it was reviewing further examples and that AI Overviews only appeared when it was very confident of accuracy. Critics argue that when it comes to health, even small errors can have serious consequences.