Google's AI Overview feature, rolled out in May 2024 after a year of limited SGE experimentation, is delivering misinformation that ranges from bizarre to perilous, particularly in public health contexts. The feature has been criticized for providing inaccurate information, such as incorrect historical facts, misleading health advice, and erroneous technical details. Examples include:
- Misleading Historical Information: AI Overview incorrectly listed US presidents attending UW-Madison, including James Buchanan and JFK, who could not have attended due to their death dates.
- Inaccurate Health Advice: AI Overview suggested dangerous health practices, such as drinking human bodily fluids to flush kidney stones and drinking mint tea for appendicitis, which could lead to serious health risks.
- Technical Errors: Incorrect definitions and advice for SEO practices were also noted.
The article highlights the potential dangers of such misinformation, especially for vulnerable individuals who might take these AI-generated answers at face value. The author argues that Google should turn off AI Overviews for medical queries to prevent harm and maintain public trust. The piece also references a theory by ex-Googler Scott Jenson, suggesting that Google's rush to implement AI features might be driven by panic rather than careful consideration.
The author calls on the SEO community to continue documenting AI Overview errors and urges Google to prioritize user safety over profit, particularly in the context of health information.
