My Month of Relying on AI Chatbots for News: A Journey Through Unreliable and Erroneous Information

A recent experiment by a journalism professor examined the reliability of generative AI systems for news reporting, especially in Québec. The study utilized various AI tools, including Google’s Gemini and ChatGPT, to gauge their accuracy in delivering factual news. Results revealed significant issues: 18% of responses lacked credible sources, with many citing fake outlets like examplefictif.ca. Only 37% of the URLs provided were valid, often leading to errors. While 47% of summaries were accurate, they included instances of plagiarism. Specific inaccuracies, such as fictionalized asylum-seeker stories and misreported events, highlighted the risks of misinformation. Overall, while AI can generate contextual insights, many conclusions drawn were unsupported or misleading. The findings align with a broader report indicating nearly half of AI-generated answers contained significant issues. Users are urged to approach AI-generated news with caution, as reliability remains a critical concern in the evolving landscape of digital journalism.

Source link