Google Expands AI Health Features: User-Sourced Advice Raises Accuracy Concerns

Google Expands AI Health Features Despite Accuracy Concerns

Google has announced plans to expand its AI-powered health information features, despite ongoing criticism over the accuracy of its current offerings. Karen DeSalvo, Google’s chief health officer, revealed that the company intends to broaden its AI Overviews to cover more health topics, while also introducing a new feature called “What People Suggest” that will provide health advice sourced from internet users.

This expansion comes amid growing concerns about the reliability of Google’s AI-generated health information. The company’s AI Overviews have been plagued by inaccuracies, including bizarre claims about baby elephants and pizza. A study conducted by Columbia’s Tow Center for Digital Journalism found that the AI’s error rate was approximately 60% for basic questions, raising alarm bells about the potential risks of disseminating incorrect health information to the public.

Google’s track record in the health sector has been mixed. The company shuttered its dedicated health division in 2021, resulting in layoffs and staff reassignments. This history has led experts to question Google’s ability to provide reliable healthcare information through AI, particularly given past failures in health tech projects.

The healthcare industry presents unique challenges that have proven difficult for big tech companies to navigate successfully. Rajiv Leventhal, a senior analyst at Emarketer, emphasized the complexities involved in disrupting the healthcare sector, suggesting that Google may face significant hurdles in its latest endeavor.

In response to criticism, Google claims to have implemented guardrails and policies to mitigate low-quality responses from its AI. The company states that it uses examples of errors to improve the AI and take corrective actions. However, Google acknowledges that a lack of high-quality web content for certain health questions can lead to incorrect AI responses.

The announcement of these expanded AI health features comes on the heels of a previous inquiry into whether Google tested an experimental AI on children, which raised serious ethical concerns.

As Google moves forward with its plans to expand AI-powered health information, the company faces the challenge of improving accuracy and reliability while navigating the complex landscape of healthcare information dissemination. The success of these initiatives may hinge on Google’s ability to address the current shortcomings of its AI systems and build trust among healthcare professionals and consumers alike.