Study Reveals AI Search Tools Frequently Provide Inaccurate Information

Research shows over 60% of AI-generated search responses are incorrect, raising concerns about reliability and impact on publishers.

A Columbia Journalism Review study found that AI search tools provided incorrect answers to more than 60% of queries when asked to identify basic details about news articles.
The study evaluated eight AI models, including ChatGPT Search and Google's Gemini, with error rates ranging from 37% to 94%.
AI tools often fabricated URLs, cited incorrect sources, or failed to provide citations entirely, undermining accuracy and publisher attribution.
Generative AI search tools frequently presented false information with unwarranted confidence, rarely acknowledging uncertainty or declining to answer.
Concerns have been raised about AI tools bypassing publisher restrictions and diverting traffic away from original content, impacting the online media economy.