ChatGPT Search Struggles with Accuracy in News Attribution, Study Finds

ChatGPT search can't find the real news, even with a publisher holding its hand

OpenAI proudly debuted ChatGPT search in October as the next stage for search engines. The company boasted that the new feature combined ChatGPT's conversational skills with the best web search tools, offering real-time information in a more useful form than any list of links. According to a recent review by Columbia University's Tow Center for Digital Journalism, that celebration may have been premature. The report found ChatGPT to have a somewhat lassie-faire attitude toward accuracy, attribution, and basic reality when sourcing news stories. What's especially notable is that the problems crop up regardless of whether a publication blocks OpenAI's web crawlers or has an official licensing deal with OpenAI for its content. The study tested 200 quotes from 20 publications and asked ChatGPT to source them. The results were all over the place. Sometimes, the chatbot got it right. Other times, it attributed quotes to the wrong outlet or simply made up a source. OpenAI's partners, including The Wall Street Journal, The Atlantic, and the Axel Springer and Meredith publications, sometimes fared better, but not with any consistency. Gambling on accuracy when asking ChatGPT about the news is not what OpenAI or its partners want. The deals were trumpeted as a way for OpenAI to support journalism while improving ChatGPT's accuracy. When ChatGPT turned to Politico, published by Axel Springer, for quotes, the person speaking was often not whom the chatbot cited. The short answer to the problem is simply ChatGPT's method of finding and digesting information. The web crawlers ChatGPT uses to access data can be performing perfectly, but the AI model underlying ChatGPT can still make mistakes and hallucinate. Licensed access to content doesn't change that basic fact. Of course, if a publication is blocking the web crawlers, ChatGPT can slide from newshound to wolf in sheep's clothing in accuracy. Outlets employing robots.txt files to keep ChatGPT away from their content, like The New York Times, leave the AI floundering and fabricating sources instead of saying it has no answer for you. More than a third of the responses in the report fit this description. That's more than a small coding fix. Arguably worse is that if ChatGPT couldn't access legitimate sources, it would turn to places where the same content was published without permission, perpetuating plagiarism. Ultimately, AI misattributing quotes isn't as big a deal as the implication for journalism and AI tools like ChatGPT. OpenAI wants ChatGPT search to be where people turn for quick, reliable answers linked and cited properly. If it can't deliver, it undermines trust in both AI and the journalism it's summarizing. For OpenAI's partners, the revenue from their licensing deal might not be worth the lost traffic from unreliable links and citations. So, while ChatGPT search can be a boon in a lot of activities, be sure to check those links if you want to ensure the AI isn't hallucinating answers from the internet.

Don't trust ChatGPT Search and definitely verify anything it tells you

Columbia's Tow Center for Digital Journalism report shows that ChatGPT Search may not be as accurate as advertised. In October, OpenAI integrated ChatGPT Search into ChatGPT, promising an experience in which users could browse the web and access the latest news from its news partners and sites that have not blocked OpenAI's web crawler. A new review by Columbia's Tow Center for Digital Journalism shows that the process may not be as efficient as it sounds. The Tow Center performed a test to determine how well publisher content is represented on ChatGPT. It selected 10 articles from 20 random publishers who partnered with OpenAI, are involved in lawsuits against OpenAI, or unaffiliated publishers who either allowed or blocked the web crawler. Also: '12 Days of OpenAI' promises product launches and demos - here's how to watch The researcher then extracted 200 quotes, which, when run among search engines like Google or Bing, pointed back to the source in the top three results. Finally, it was time to let ChatGPT identify the quotes' sources. Ultimately, the goal was to see if the AI accurately serves publications, giving them credit for their work. If the approach worked as advertised, it should be able to attribute the sources just as well. The results varied in accuracy, some entirely correct or incorrect, and some partially correct. Yet, nearly all answers were presented confidently, without the AI saying it couldn't produce an answer even from publishers who had blocked its web crawler. Only in seven of the outputs did ChatGPT say to use words or phrases that insinuated it was unclear, as seen below: "Beyond misleading users, ChatGPT's false confidence could risk causing reputational damage to publishers," the article stated. Also: This new AI podcast generator offers 32 languages and dozens of voices - for free That statement was backed up by an example in which ChatGPT inaccurately attributed a quote from the Orlando Sentinel to a Time article, with over a third of ChatGPT's responses with incorrect citations being of that nature. In addition to harming traffic, misattribution can harm a publication's brand and trust with its audience. Other problematic findings from the experiment include ChatGPT citing an article from The New York Times, which has blocked it, from another website that had plagiarized the article, or the citing of a syndicated version of a piece from MIT Tech Review instead of the original article, although MIT Tech Review does allow crawling to take place. Also: '12 Days of OpenAI' promises product launches and demos - here's how to watch Ultimately, this research points to a larger question of whether or not partnering with these AI companies offers publishers more control and whether creating new AI search engines truly benefits publishers or hurts their businesses in the long run. The data behind the methodology is shared on GitHub and can be looked at by the public. Consumers should always verify the source by clicking on the footnote the AI provides or doing a quick search on an established search engine, such as Google. These extra steps will help prevent hallucinations.