A joint investigation by The New York Times and AI company Oumi found that Google’s AI-generated search summaries have a 91 percent accuracy rate, a figure that suggests tens of millions of incorrect answers are served to users hourly and raises questions about the reliability of the company's flagship AI product.
The findings, based on a joint test of Google's AI summaries, highlight the scale of the challenge facing Alphabet Inc. (GOOGL) as it integrates generative AI into its core search engine. While 91 percent accuracy appears high, the sheer volume of queries Google processes—estimated at 5 trillion annually—means a 9 percent error rate can result in a significant volume of misinformation.
The investigation noted that the problem extends beyond factual errors. Over half of the AI-generated answers that were deemed correct provided citation links that did not actually support the conclusion, undermining the sourcing and verification process for users.
This report could create a negative feedback loop for Alphabet, potentially impacting its stock as investors weigh the reputational damage and the cost of remediation. The findings open the door for competitors like Perplexity AI and OpenAI, which are developing their own AI-native search solutions, to capture market share from users seeking more reliable information. Addressing these accuracy issues will likely require significant additional investment in research and development, potentially affecting Google's short-term profitability.
This article is for informational purposes only and does not constitute investment advice.