A recent study by the Tow Center for Digital Journalism has found that OpenAI’s AI chatbot, ChatGPT, consistently produces inaccurate citations for publishers’ content, regardless of whether they have a licensing deal with the company. The researchers tested 200 block quotes from various publishers and found that ChatGPT returned partially or entirely incorrect responses on 153 occasions, often without acknowledging its inability to provide accurate information. The study highlights several concerns, including the potential for unreliable sourcing, decontextualized data, and inconsistent responses, which may lead to reputation risks and commercial losses for publishers.
Source: https://techcrunch.com/2024/11/29/study-of-chatgpt-citations-makes-dismal-reading-for-publishers