Over 60% of AI chatbot responses are incorrect

A study by Columbia Journalism Review's Tow Center for Digital Journalism found that over 60% of responses from popular AI chatbots are incorrect or misleading. The chatbots often fail to credit original news sources properly. Researchers tested eight AI chatbots, including ChatGPT and Perplexity, and found that many provided wrong answers confidently. They frequently linked to broken URLs or unauthorized content, undermining publishers' traffic and revenue. The study highlights a growing issue for news organizations as AI tools replace traditional search engines. Publishers struggle to monetize their content, and misattributions can damage their reputations.