AI search engines provide incorrect answers over 60% now

A study by Columbia Journalism Review found that AI search engines provide incorrect answers over 60% of the time. This raises concerns about their reliability, especially as about 25% of Americans now use these tools instead of traditional search engines. The research tested eight AI-driven search tools, revealing significant error rates. For instance, ChatGPT Search misidentified 67% of articles, while Grok 3 had a 94% error rate. Premium versions of some tools performed worse in certain cases. Additionally, the study noted that some AI tools ignored publisher requests to restrict access to their content. For example, Perplexity's free version accessed paywalled National Geographic articles despite explicit prohibitions.