Google’s new AI review is being criticized for generating weird and incorrect answers, such as suggesting users put glue on pizza or eat rocks. This has led to Google manually disabling AI insights for specific searches.

AI experts, including Gary Marcus, say that achieving high accuracy in AI responses is difficult and may require advanced reasoning. Google’s current AI models, such as Gemini, have struggled with quality control, tarnishing the company’s reputation.

The pressure to compete with competitors like Bing and new AI startups is leading to premature releases of AI features.

Source: The Verge