AI-powered search engines had promised an accelerated, improved way of finding information. Yet while applications like ChatGPT Search, Perplexity, Gemini, and Grok are dominating, the reality is evident: these chatbots aren’t simply making citations wrong—they’re actually killing off the open web.
The Tow Center for Digital Journalism at Columbia University recently published a report which found that AI search engines could not locate the correct citation more than 60% of the time. Others, including Grok-3 Search, fabricated links 94% of the time, typically directing individuals to dead pages.
Bad citations are just part of the story. AI chatbots are basically redefining the manner in which people consume information, eroding public trust in journalism and cutting traffic that publishers rely on to survive.
Fake Links, False Confidence, and a Citation Crisis
The Tow Center test compared eight AI search engines on 200 queries and asked them to locate article title, date, publisher, and URL. And what they discovered was surprising:
- Grok-3 and Gemini fabricated more fake links than they returned correctly.
- ChatGPT provided false citations 134 times—and only qualified the accuracy in only 15 instances.
- Perplexity was most accurate, but still made a mistake 37% of the time.
Beyond the inaccuracies, the bigger issue is confidence. AI chatbots, as opposed to older search engines, release information as absolute fact, not something users are easily able to discern credibility of. A hallucinated link provided in surety is possibly worse than a direct omission.
AI Search Drives 96% Less Traffic Than Google—A Death Blow to Digital Publishing?
Even when AI search engines accurately attribute sources, they don’t send users to them. In an experiment, TollBit found that AI responses send 96% fewer referral visits than traditional Google search.
Similarly, Northwestern University’s Generative AI in the Newsroom initiative reviewed five months of ChatGPT and Perplexity data and found that news sites received just 3.2% of ChatGPT’s filtered traffic and 7.4% of Perplexity’s traffic.
This is disastrous for publishers. News organizations rely on search engines to deliver clicks, ad revenue, and subscriptions. If AI search engines dominate but don’t refer users to original sources, journalism itself is at risk.
What Comes Next? The Fight for AI-Driven Visibility
The media industry has been pushing back since AI search engines have been failing on accuracy and traffic. Some of the key debates include:
- Should AI companies pay publishers? Some are for a revenue-sharing model that is similar to how Google compensates news publishers.
- Should AI Search be regulated? There might be an urgency for regulatory intervention due to the rampant misinformation carryover by chatbots.
- Will there be a change in SEO? New strategies on how to make their content favorable to AI would need to be established by publishers so they get recognized appropriately.
Conclusion: AI Search Needs a Fix—Fast
AI search engines were supposed to democratize information access, but now they are killing it. If chatbots do not make citation easier and correctly direct people to primary sources, then the internet is likely to be turned into an in-cycle misinformation system.
The war for moral AI search has begun. If AI companies are going to lead the future of search, they need to start by fixing their most basic function: getting the facts right.
Follow us on WhatsApp, Telegram, Twitter, and Facebook, or subscribe to our weekly newsletter to ensure you don’t miss out on any future updates. Send tips to editorial@techtrendsmedia.co.ke
