Artificial intelligence is about to enter the search business. Or so we are told. As Google seems to be getting worse, and tools like ChatGPT, Google Gemini, and Microsoft Copilot seem to be getting better, we seem to be struggling to find a new way to find and consume information online. Companies like Perplexity and You.com are positioning themselves as the next generation of search products, and even Google and Bing are betting that AI is the future of search. Goodbye, 10 blue links; hello, direct answers to all my weird questions about the world.
But what you have to understand about search engines is that search engines are a lot of things. For all the people who use Google to find important and hard-to-access scientific information, many more use it to find their email inboxes, visit the Walmart website, or remember who the presidents were before Hoover. And then here’s my favorite fact: Every year, tons of people visit Google and type “google” into the search box. We think of Google primarily as a research tool, but in reality, it’s asked to do anything you can think of, billions of times every day.
The real question facing all these would-be Google killers, then, isn’t how well they find information. That’s how well they can do everything Google does. So I decided to put some of the best new AI products to the real test: I took the latest list of Google’s most searched queries and questions, based on SEO research firm Ahrefs, and plugged them into various AI tools. In some cases, I’ve found that these language model-based bots are indeed more useful than Google results pages. But for the most part, I see how difficult it will be for anything—AI or otherwise—to displace Google at the center of the web.
People who work in search always say that there are basically three types of queries. The first and most popular is navigation, which allows people to access a website simply by typing its name. In fact, all of the most popular queries on Google, from “youtube” to “wordle” to “yahoo mail”, are navigational queries. In fact, this is the main job of search engines: to get you to a website.
In fact, a search engine’s main job is to get you to a website
For navigational queries, AI search engines are generally worse than Google. When you do a navigational Google search, it’s rare that the first result isn’t what you’re looking for — and, of course, showing you all of this when all Google should actually be doing is taking you directly to Amazon The result is very strange. com or whatever, but it’s fast and rarely goes wrong. The AI bot, on the other hand, likes to think for a few seconds and then provide a bunch of quasi-useful information about the company, when all I want is a link. Some don’t even link to amazon.com.
I don’t hate the extra information, I hate how long These AI tools work for my needs. Waiting 10 seconds to generate three paragraphs of text about Home Depot is not the way to go. I just want the link to Home Depot. Google wins this game every time.
The next most popular type of search is an information query: you want to know something specific, and have a right answer to it. “NFL scores” is a very popular information query; “What time is it” is another; so is “weather.” It doesn’t matter who tells you the score, the time, or the temperature, it’s just something you need to know.
Here, the results are all over the map. For real-time information like sports scores, AI isn’t trustworthy: Both You.com and Perplexity frequently gave me outdated information, although Copilot usually got it right. Not only does Google do it right, but it usually pops up a widget with additional statistics and information that’s better than the other widgets. The same goes for anything that requires your specific location or background—Google may have information about you, but AI bots mostly don’t.
When it came to more commonly occurring information like “how many weeks are there in a year” or “when is Mother’s Day”, everything I tested was correct. In many cases, I actually prefer the AI’s answer, which adds some useful context. But I’m not sure how often I can trust them. Google tells me there are 52.1429 weeks in a year, but You.com explains that it’s actually 52 weeks and one day, plus one day in a leap year. This is more useful than 52.1429! But then Perplexity told me that in fact, there are 52 weeks in an ordinary year and 52 weeks and one day in a leap year – and then the two sentences directly contradicted each other. Here is the complete answer; just trying to understand it:
There are approximately 52 weeks in a normal year, and 52 weeks and 1 day in a leap year. To be more precise, a normal year actually consists of 52.143 weeks, which means there is one extra day in a normal year. On the other hand, with certain exceptions, leap years occur every four years and have 52 weeks and 2 days. The difference in week numbers is due to an extra day in normal years and an extra day in February in leap years.
After more research, I am now convinced that the answer is what You.com says it is. But it all took too long and forced me to fact-check my searches, which kind of defeats the purpose of helping me summarize things. Google continues to win here, and for one thing only: speed.
However, the opposite is true for one subtype of information inquiry. I call them hidden information queries. The best example I can provide is the very popular query, “How to take a screenshot on Mac”. There are a million pages on the internet with the answer – just Cmd-Shift-3 to take up the whole screen, or Cmd-Shift-4 to capture the selection, you’re welcome – but this information is often buried beneath a lot of advertising and SEO Rubbish. All the AI tools I’ve tried, including Google’s own search generation experience, just scrape this information and serve it directly to you. it’s great!
Are there complex issues that threaten the network’s business model and structure? Yes! But as a pure search experience, it’s much better. I’ve gotten similar results, asking about ingredient substitutions, coffee ratios, headphone waterproof ratings, and any other information that’s easy to know but often hard to find.
This brings me to the third type of Google search: Discovery queries. These questions have no single answer, but are the beginning of a learning process. On the most popular list, things like “How to tie a tie,” “Why was the chainsaw invented,” and “What is TikTok” were all considered exploratory queries. If you’ve ever Googled the name of a musician you’ve just heard, or looked up something like “things to do in Helena, Montana” or “NASA history,” you’re exploring. According to the rankings, these are not the main reasons people use Google. But this is where artificial intelligence search engines shine.
Like, wait: Why? yes Was the chainsaw invented? The First Officer gave me a multi-part answer to their medical origins before describing their technological evolution and eventual adoption by lumberjacks. It also gave me eight very useful links to read more. Perplexity gave me a much shorter answer, but also included some cool pictures of old chainsaws and a link to a YouTube explainer on the subject. Google’s results included a lot of the same links, but didn’t do any synthesis for me. Even its generated search only gave me the bare minimum.
My favorite thing about AI engines is quotes. Perplexity, You.com and other sites are slowly getting better at linking to their sources, often inline, which means if I come across a particular fact that piques my interest, I can go straight from there to the source. They don’t always provide enough resources, or put them in the right place, but it’s a good and useful trend.
One of my experiences while doing these tests was actually the most eye-opening. One of the most searched questions on Google is simple: “what to look at”.Google has a whole specific page design for this, with rows of posters saying “Preferred” e.g. Dune: Part 2 and imaginary; “For you”, to me includes dead Pool and stop and catch fire; Then there are options for popular titles and sorting by genre.No AI search engine does this well: Copilot lists five popular movies; Confusion offers a handful of seemingly random options girl 5eva arrive hunt arrive general; You.com gave me a bunch of outdated information and suggested I watch “The 14 Best Netflix Original Movies” without telling me what they were.
Artificial Intelligence is the right idea, but chatbots are the wrong interface
In this case, AI was the right idea—I didn’t want a bunch of links, I wanted answers to questions—but the chatbot was the wrong interface. The same goes for search results pages, for that matter! Google obviously realized this was the most asked question on the platform and was able to design something that worked better.
In a way, this is a perfect summary of the current situation. For at least some web searches, generative AI may be a better tool than the search technology of decades past. But modern search engines are about more than just linking pages. They are more like micro-operating systems. They can answer questions directly, and they have calculators, converters, flight selectors, and a variety of other built-in tools to get you where you want to go in just a click or two. According to these charts, the goal of most search queries is not to start a journey of information wonders and discovery. The goal is to get a link or answer, and then leave. Currently, these LLM-based systems are too slow to compete.
I think the biggest problem is not the technology, but the product. Everyone, including Google, believes AI can help search engines better understand questions and process information. This is the current industry consensus. But will Google be able to reinvent its results pages, business model, and the way it presents, summarizes, and presents information faster than AI companies can transform their chatbots into more complex, multifaceted tools? The ten blue links are not search answers, but neither is the generic text box. Search is everything, everything is search. It will take more than chatbots to kill Google.