Google's AI Overview seems to be spewing inaccurate, dangerous answers

Google's new "AI Overview" feature, which aggregates AI-generated responses above search results, has been criticized for providing inaccurate and dangerous advice. This includes a bizarre suggestion to mix glue with cheese and a hazardous cleaning tip that could produce mustard gas. The feature, aimed at competing with similar offerings by Microsoft and OpenAI, is also accused of blatant plagiarism, adding personal touches to content sourced from other sites without permission.

Concerns are growing that this AI tool, by presenting summaries, may reduce traffic to the original content creators, potentially harming those sites financially. Critics, including the News Media Alliance, argue for legal measures against AI systems that use publisher content without compensation. Google maintains that the feature helps users explore a broader range of websites, but includes a disclaimer about the experimental nature of its AI.

Google's "AI Overview" feature, which aggregates results compiled by artificial intelligence, has been accused of plagiarism due to its method of integrating content from other websites into its responses2. Social media users have highlighted instances where the language used by Google closely mirrors that found on original websites, without adequate attribution2. For example, Google added the phrase "my kid’s favorite" to a smoothie recipe, which was lifted from a recipe website that had a similar description5. This practice of borrowing language directly from sources, with minimal or no modification, raises concerns about the ethical use of content and the potential for infringing on the intellectual property of the original content creators. Critics argue that this not only misrepresents the originality of the content provided by Google's AI but also undermines the traffic and revenue of the websites that originally published the content.