Google Announces Improved AI Overviews for Health Guidance Despite Previous Errors
Google’s AI and Health Information: What You Need to Know
Concerns Over AI Accuracy
There has been growing concern regarding the reliability of Google’s AI, particularly its so-called "AI Overviews." Reports indicate that these overviews have repeatedly provided information that is not only incorrect but potentially dangerous. Recently, Google introduced a feature that allows the AI to offer health advice, which raises alarms about the accuracy of the information users might receive.
New Health-Centric Features
In a recent blog post, Google’s Chief Health Officer Karen DeSalvo highlighted enhancements to the Gemini AI model, claiming that it would now cover a broader range of health topics. However, one concerning addition is the "What People Suggest" feature, which offers health advice sourced from amateur internet users. While this feature is reportedly available on mobile devices in the US, many users have found it hard to access it through Google’s existing platforms.
Track Record of Inaccuracy
The track record of Google’s AI Overviews is troubling. Various sources point out that the AI has produced some absurdly incorrect information, such as claiming that baby elephants can fit in a human hand and suggesting illogical actions, like putting glue on pizza. A study from Columbia University’s Tow Center for Digital Journalism found that Google’s Gemini chatbot, which backs the AI Overviews, answered basic questions incorrectly 60% of the time. Despite this, Google has not publicly disclosed plans to address these significant inaccuracies.
A Shift Away from Human Expertise
Interestingly, Google once employed hundreds of experts focused on health information. However, in 2021, the company decided to shut down its health division, leading to layoffs and a reorganization of the remaining staff. Experts had previously expressed concerns about Google’s commitment to providing reliable health information, especially in light of past failures in health tech projects. Many industry analysts remain skeptical about Google’s ability to deliver high-quality healthcare data through AI, given the complexity of the sector.
The Challenge of Trustworthy Information
If Google aims to transform the healthcare industry using AI, it faces unique challenges. As noted by industry analyst Rajiv Leventhal, disrupting healthcare is no simple task. Effective healthcare requires accurate and trustworthy information, and that is where Google’s current AI shortcomings become particularly worrying. Questions about how Google will ensure the accuracy of health-related content remain unanswered.
Google’s Response to Criticism
In response to inquiries about their AI Overviews, Google has suggested that the lack of high-quality web content for certain questions might be a reason for the inaccuracies. They claim to have policies in place to mitigate low-quality responses and state that they incorporate issues that arise into future improvements. A spokesperson mentioned that the company will continue refining their systems as they receive feedback.
The Future of AI in Healthcare
While the potential for AI to revolutionize healthcare is significant, the existing issues with accuracy must be addressed to build trust. As of now, users should exercise caution when relying on Google’s AI for medical advice until significant improvements in information reliability are made.
