Google's AI Overviews feature, powered by Gemini large language models, has faced criticism for providing inaccurate and potentially harmful search results.

Following backlash, Google admitted to mistakes in misinterpreting queries and relying on unreliable sources like Reddit.

The company has implemented enhanced guardrails for critical topics such as news and health, limiting satire and user-generated content, and refining triggers for AI Overviews.

Users now have the option to disable the feature to mitigate misinformation risks.

Despite efforts to enhance accuracy, critics warn against relying solely on AI for search results.

Google plans to expand the feature to over a billion users by the end of the year and is considering monetizing it with 'Sponsored' ads.

The company has acknowledged the need for responsible AI deployment and has made organizational changes to prioritize accuracy, trustworthiness, and transparency in its AI products.

Google has defended its AI overviews feature but acknowledged some answers were inaccurate or harmful, including dangerous advice and conspiracy theories.

The company has implemented fixes to prevent such errors, including limiting user-generated content and improving detection mechanisms for nonsensical queries.

AI experts have warned against relying solely on AI-generated answers, citing the potential for bias, misinformation, and hallucination.