Google is facing its most severe quality and trust crisis since the restructuring of its search experience. Although "AI Overviews" have become the core of Google Search, frequent "hallucinations" - including fabricated facts, contradictions, and providing incorrect advice in critical areas such as healthcare - have forced the tech giant to take emergency measures.
According to recent job postings, Google is hiring "AI Answer Quality Engineers," who are specifically responsible for optimizing AI models and generative answers under the Search Results Page (SRP). The job requirements indicate that these engineers will work on improving AI performance for complex queries, ensuring accuracy in information while expanding infrastructure. This is seen as Google's first indirect admission that its AI Overview feature has significant reliability issues.
Actual tests have revealed the severity of the problem: when asked about the valuation of the same startup company twice, the AI Overview provided completely wrong reference answers ranging from $4 million to $70 million. More seriously, The Guardian recently disclosed that Google's AI gave health advice for serious diseases like pancreatic cancer that was completely opposite to medical standards and even potentially life-threatening.
In addition, Google's recently tested AI rewriting news headline feature in the Discover feed has been strongly criticized by publishers for generating misleading "clickbait" content. As users' natural trust in Google's search results is being challenged, eliminating AI hallucinations has become a crucial battle for Google to maintain its dominance in search.
