Survey: 83% of users prefer AI search over ‘traditional’ Googling

Quoted in the article

The days of ‘Googling’ may be numbered. We asked Innovating with AI readers about their experience with AI search solutions, and more than 83% said they found AI-powered tools more efficient than traditional search.

With the emergence of a new wave of AI search tools, an increasing number of consumers are shifting away from the traditional search experience.

According to data collected by Statcounter, Google’s global market share fell below 90% for the first time since 2015 in October 2024. While there are other factors that could cause such a drop, it is likely that the growing popularity of AI-driven search tools like ChatGPT, Grok, and Perplexity AI are at least partially to blame. 

But why is AI search gaining so much momentum? For many users, the answer is that it provides answers much more quickly than traditional search. Rather than sifting through pages of hyperlinks and meta descriptions, you can simply read a single response that collates resources into a single response. But in practice, things are a little more complicated. 

Entering the Era of AI Search 

With ChatGPT reaching 200 million weekly active users at the end of 2024, it’s clear that the way human users search for information is changing. Innovating With AI (IWAI) conducted a survey of frequent AI users and found that nearly all of them—91%—utilize popular large language models (LLMs) such as ChatGPT for their internet searches.

We’ve also seen leading search providers, such as Google and DuckDuckGo, start to pivot toward generative AI-driven user experiences, most notably by adding AI-generated summaries to the top of search results. And as mentioned above, AI can provide users with answers more quickly than the traditional search experience, leading some to refer to ChatGPT as a “Google Killer.” 

However, Nick Reese, an Adjunct Instructor at NYU’s Center for Global Affairs, who also served as the first ever Director of Emerging Technology Policy at the US Department of Homeland Security, suggests that the idea of AI search replacing traditional search is a false dichotomy. 

“I think that those tools are building upon traditional search, and will present an evolution of search. And so when we talk about replacing that, that implies a dichotomy of, we’re going to shift from one thing entirely to another,” Reese said. “I think it would be more accurate to say that if you look at search as kind of a singular pursuit, we’re seeing an evolution in that, and I do think that that evolution will become the preferred tool for executing search in the near future.” 

Many Users Are Prioritizing Efficiency

Reese also suggests that AI search has several key selling points that drive adoption. “The advantage with the new AI-enabled search results, or search products, is that they give you results in plain language,” he says. “They can summarize things for you, give you links if you need more information, and provide good citations. They don’t just give you a link; they actually answer your question.”

With platforms like Perplexity AI or ChatGPT, users don’t need to click through to third-party sites for an answer; they can receive a cohesive response with a single click. They also don’t have to worry about navigating search results that have been crammed with SEO-driven content written by algorithms rather than people. 

Of course, there are still areas of search where tools like Google and Bing still have an advantage. Reese argues that traditional tools are good for looking at news stories and current events, whereas AI search tools are better at responding to queries when the user is unsure what they’re looking for (potentially exploring new topics and concepts). 

I’d suggest search engines might have the edge for browsing the news, because users can easily
scroll through a list of reputable sources covering a recent news story with a much higher degree of choice than a tool like Perplexity or ChatGPT. Likewise, an LLM would potentially save you time by offering a summary about an unfamiliar topic more quickly than if you had to track down another website’s article or blog post.

However, AI search’s ability to provide users with quick answers comes at a significant cost; those answers are often incorrect, a phenomenon known as AI “hallucinations.” This suggests that traditional search may have some sticking power, at least for the foreseeable future. 

The Problem with AI Search

While AI search has undoubtedly gathered significant momentum, many users continue to gravitate toward traditional search experiences. Research from link-building provider OneLittleWeb found that between April 2024 and March 2025, Google received 26x more traffic per month compared to ChatGPT. 

The same report also found that while AI chatbots experienced an 80.92% increase in traffic, search engine traffic remained stable, with a total drop in visits of less than 1%.

Other research, conducted by YouGov, suggests that not all users are embracing AI search. Only 38% of users surveyed reported reading AI-generated search summaries. 

Part of the reason why some are ignoring AI search is the sentiment that the responses of LLMs can’t be trusted. LLM-powered tools like ChatGPT and Perplexity AI utilize language models to predict outputs in response to user questions, resulting in responses that are often verifiably incorrect. 

Hallucinations Abound

The New York Times reports that o3, one of OpenAI’s most powerful models, hallucinated 33% of the time when performing the PersonQA benchmark test, a set of questions and facts designed to help measure the accuracy of language models. That’s a 1 in 3 chance that the response you receive is going to be completely false. 

We’ve also seen well-publicized examples of Google AI claiming that the current year is 2024 rather than 2025, and even encouraging users to add glue to pizza.

These hallucinations mean that users either have to use AI search for “low stakes” searches, fact-check the output against a third-party source, or just accept the risk of being misinformed. For this reason, many users will stick to traditional search, as it’s easier to get reliable answers from trustworthy sources.

That being said, hallucinations could become less prevalent over time as vendors adopt methods such as reinforcement learning from human feedback (RLHF) and chain of thought reasoning to improve the overall reliability of responses. For now, having the LLM provide citations and in-response links serves as a temporary solution.

Google’s Move to Cannibalize Human Search 

When considering the future of search, it is difficult to overlook the role of Google, which has been moving toward cannibalizing traditional search by ranking its own AI-generated responses above human-written content from third-party sites. 

Chris Andrew is the CEO and co-founder of Scrunch AI, a startup helping organizations to rank in AI search. He suggests that Google is “accelerating the cannibalization of traditional search because humans prefer the AI overview,” citing Google AI overviews 1.5 billion monthly active users. 

“[Google is] challenging [its] own business model as they figure out new ways to monetize AI overviews. But I think what we will see is…. AI Mode, the new capabilities Google announced last week, [becoming] the dominant search experience. It might not happen overnight, but we will see a world where the traditional kind of search results will get pushed further down or become completely irrelevant,” Andrew said. 

We’re entering a situation where Google is promoting the responses of its own proprietary AI model over those of third-party voices, which is a departure from the traditional search model, where users were connected with a high volume of third-party sources. 

There are serious ethical concerns about this, particularly when the company’s own AI has been trained on that content without compensation or permission, and has developed a competing product that directly undermines its traffic. 

For instance, France’s competition watchdog has fined Alphabet $271.7 million for training its AI on content taken from publishers and news agencies without notifying them. 

The End of the Information Age 

Although it appears that traditional search will remain around for a while longer, Google’s actions to promote AI-generated content over human-written content suggest that search as we know it is going to change. My assessment of what’s happening now is that Google is promoting its proprietary AI overviews at the expense of third-party publishers, whose results are pushed further down in search rankings while the platform blurs the lines between being a content provider and a search engine.

You could argue that AI overviews offer users a better user experience by providing them with quicker answers to questions, but given the prevalence of hallucinations and the decline in visibility of third-party sites, the decline of traditional search would be a significant blow to freedom of information. 

If search providers continue to phase out human-written content, the average internet user will have less freedom to access information independently, even if AI search providers offer citations. 

Case in point: Asking Google, “What caused the Second World War?” yields 15 pages of results. Asking the same question to AI Mode generated just 12 links. For this query, AI search offered fewer choices and less control. 

The writing on the wall is clear: Brands and publications will have to work much harder to reach their audiences if traditional search continues to be cannibalized by AI-centric platforms. Whether that comes down to optimizing content to rank in AI search with generative engine optimization (GEO) or doubling down on social media marketing, it’s likely that organic search traffic for most companies will never be the same again. 

Read the article here.