Artificial intelligence tools are becoming increasingly capable, but just how effective are they at searching the web? It’s a critical question, especially considering that nearly a third of U.S. users now rely on AI instead of traditional search engines like Google, according to research from Future (the publisher of TechRadar).
Some users turn to AI chatbots like ChatGPT, while others prefer AI tools specifically designed for research, such as Perplexity. Even those who stick to Google are still interacting with AI-powered results through Google’s automated summaries. In short, AI-driven search is everywhere. But does that mean it’s reliable?
AI vs. Google: Which One Wins at Search?
Just because people are using AI search doesn’t mean it’s always the best option. To put AI search capabilities to the test, we examined four leading AI chatbots—OpenAI’s ChatGPT, Google’s Gemini, Anthropic’s Claude, and Perplexity AI—to see how well they retrieved and summarized information.
The results were mixed. While none of the tools completely failed, they often struggled with accuracy, and their summaries were frequently unclear or misleading. A broader study by the Tow Center for Digital Journalism, reported in the Columbia Journalism Review, reinforced these findings. The research examined eight major AI models, including ChatGPT, Perplexity, Copilot, Grok, and Gemini, and found systemic flaws: “confident presentations of incorrect information, misleading attributions to syndicated content, and inconsistent information retrieval practices.”
Across all models, over 60% of AI-generated responses contained inaccuracies. Perplexity performed the best, but it still got 37% of answers wrong. At the other extreme, Grok delivered incorrect information 94% of the time—an astonishingly poor result.
The Biggest Problem: AI Search Without Proper Sources
One of the most fundamental flaws of AI search engines is how they package and present information. Even when AI models don’t outright fabricate answers, they often reframe content in misleading ways. Traditional search engines act as intermediaries, directing users to original sources. AI models, on the other hand, synthesize and summarize content without always crediting the original publishers.
The Tow Center highlighted this concern:
“These chatbots’ conversational outputs often obfuscate serious underlying issues with information quality. There is an urgent need to evaluate how these systems access, present, and cite news content.”
A key problem is the lack of proper citations. ChatGPT, for example, frequently links to irrelevant sources, sends users to generic homepages, or omits citations altogether. This presents two major issues:
- Publishers lose traffic. AI models pull information from news sources but often fail to credit them, reducing engagement and revenue for the original creators.
- Fact-checking becomes difficult. Without direct access to sources, users must manually verify AI-generated responses—often by searching Google anyway.
Is AI Search Really Worth Using?
AI search tools are evolving rapidly, and some, like Perplexity, are more reliable than others. However, even the best-performing AI models require human oversight. Given their current limitations—high error rates, lack of proper citations, and frequent misinformation—AI chatbots simply aren’t ready to replace traditional search engines.
For now, if you need accurate, verifiable search results, Google (or another established search engine) remains the best choice. AI search has potential, but until it can consistently provide reliable and well-sourced answers, it’s not quite ready to take over.
