The Rise of AI Search Engines: Should Google Be Worried?

Have you ever caught yourself asking ChatGPT to search something instead of typing it into Google? I have – and it felt oddly exciting. A few years ago, Google was the uncontested king of finding stuff online. We “Googled” everything: directions, recipes, even random trivia ("Who really wrote the books of C.S. Lewis?" – yes, I got lazy). Google handled trillions of queries effortlessly. In 2024 alone it processed over 5 trillion searches, roughly 93% of the market . It seemed unstoppable. Then ChatGPT arrived on the scene, seemingly overnight, and made us question if the search game was changing forever. Today, clever new AI tools answer questions in full sentences. They summarize answers and chat with us, rather than just giving us a list of blue links to click. It’s a bit like trading a stack of encyclopedias for a very smart friend who can skim them for you.

From PageRank to AI: A Search Odyssey

Once upon a time, before Instagram or TikTok had our full attention, we had search engines. I still remember AltaVista and Lycos, the dinosaurs of the web, spitting out results that were often confusing or outdated. Then came Google in the late 1990s with its PageRank magic – finally relevant search results! It was like finding a genie in a lamp. By the 2000s Google had captured virtually all of us; as Google’s own numbers show, the typical person was happily clicking on Google 7–8 times per day to get things done.

Over the years Google kept improving. It learned to suggest searches, answer questions right at the top, and even translate languages on the fly. It built sprawling “knowledge graphs” of facts, and developed its own AI brain (DeepMind, Google Brain, LaMDA, etc.) behind the scenes. As Google’s VP of Search Elizabeth Reid puts it, they “meticulously honed” their systems to help you “find the best of what’s on the web” in the blink of an eye. I still marvel at how Google can answer “What’s the weather in Tokyo?” instantly or pull up directions faster than I can tie my shoes. For decades, Google was synonymous with search.

But change was afoot. In late 2022, OpenAI’s ChatGPT exploded into the public eye, and people were thunderstruck by how it could write essays, code, or even chat about philosophy. Many of us, including the Google team, realized: people loved the conversational, free-form way of getting answers. Soon, ChatGPT and similar tools started to dabble in search-like tasks. Instead of listing links, they summarized information in plain English. This shift was huge – like switching from reading a dictionary entry to asking a tutor directly.

According to industry insiders, OpenAI even launched something called SearchGPT in mid-2024, and rolled out a new “ChatGPT Search” feature shortly after. These tools don’t just give you URLs; they write you a short answer and point you to sources. Meanwhile Microsoft updated Bing to use ChatGPT, Google beta-launched Bard (now powered by its new “Gemini” model), and smaller startups like Perplexity.ai and You.com arrived with their own AI-driven search. In short, generative AI is now in play on every front.

Chatbots in the Driver’s Seat

So what is an AI search engine? In practice, it’s a chatty interface on top of a brainy language model. Instead of typing a query and scanning links, you ask a question and talk to the AI, as if it’s a very knowledgeable but opinionated librarian. For example, using ChatGPT or Perplexity to plan a vacation yields not just facts (“The Eiffel Tower is in Paris”) but helpful narratives (“Your best bet is to buy tickets online for the Eiffel Tower; consider going early to beat crowds”). The AI even says things like “I suggest…”, making it feel more like a friendly helper.

A recent industry survey confirms this trend. It found that 62% of people already use an AI chatbot every day, and about half of chatbot users plan to use them even more. Another survey by marketing experts found that nearly three-quarters of searchers actually use Google’s new AI summary (“AI Overview”) when it appears. People seem to love getting a quick summary at the top of search results – it saves time. (I’ll admit, I often skim Google’s AI box to see if my answer is already there before clicking any links.)

At the same time, younger users are swapping queries for conversations. YouTube CEO Neal Mohan recently observed that Gen Z often says “Not sure, ask [Bard/ChatGPT/etc.]” instead of Googling. It’s true in my own circle: I’ve seen my niece ask Bing Chat to help with homework instead of searching, because it feels more immediate (“Hey Bing, explain photosynthesis like I’m 5,” rather than Googling “photosynthesis definition”).

Yet not everyone is convinced AI chat will topple Google. A SparkToro analysis compared search volumes and found that even with ChatGPT’s explosive popularity, Google is still performing about 373 times more searches than ChatGPT. Google handled over 5 trillion queries in 2024 – and that was up 21% from 2023. By contrast, ChatGPT’s “search-like” prompts were a tiny fraction (<1%) of total query volume. In other words, most people worldwide are still typing into Google’s search box, not asking ChatGPT. Google’s head Sundar Pichai himself noted that people who try Google’s AI Overviews actually search more, not less.

So the data is mixed. On one hand, AI chat tools are wildly popular and reshuffling how we get answers. The old “deck” of search has been reshuffled, as one analyst put it. But on the other hand, statistically Google’s dominance hasn’t crumbled (yet). As OpenAI CEO Sam Altman candidly told U.S. senators, “I don’t think [ChatGPT] will replace Google as the primary search engine”. He calls Google “a ferocious competitor” with a massive AI team still making great progress. So, users love the convenience of chatbots for some tasks, but they also recognize Google’s breadth and reliability.

Google Strikes Back with AI

Google isn’t sitting quietly by either. It’s thrown everything into making its own AI-powered search magic. Remember Bard? Google launched Bard in early 2023 as an experiment, and quickly upgraded it under the “Gemini” umbrella – their latest and greatest AI model. At Google I/O 2024, they unveiled a slew of new features: AI Overviews in search, multi-step “Follow-Up” answers, even the promise of turning search results into little app-like cards. As Google’s VP of Search explained, with generative AI Google can “take the legwork out of searching” by letting you ask complex questions in one go.

According to Google, these AI features are already being used billions of times. They’ve tested it in Search Labs and found that when AI Overviews appear, users click the included links more often than before. That’s their data, anyway – Google loves to emphasize that it’s still sending traffic to websites. They’re also careful to show ads even in AI modes, to keep their business model humming.

From a user perspective, Google’s AI mode still feels like Google. You might see a block of text (like an answer paragraph) but it’s followed by source links. For example, I tried it: Google’s AI answered my question about “easy vegetarian dinners” with a nice summary, and then listed recipes from real sites below. The “assistant” text was there, but you still needed to scroll for actual pages. Sundar Pichai made this clear: Google will still point you to existing content. In Wired, he said that even with the new SGE (Search Generative Experience), Google will give you a set of cited sites so users are “consuming those sites”. The early tests seem to confirm this approach.

On the ad and SEO side, though, people are nervous. The same OrbitMedia study I mentioned found that Google’s new AI boxes can reduce clicks on websites by up to 70%. That’s understandable: if you can get a quick answer from an AI snippet, why scroll? Many web publishers worry about losing traffic. Google counters that most of the lost clicks were for queries where people just wanted a quick fact or definition anyway. The search giant keeps adding features like “People Also Ask” boxes and interactive carousels, so in some ways it already does instant answers – now they’ve just leveled it up with chatty AI.

So far Google’s bet is that it can blend the best of both worlds: fast AI answers and traditional links. As Pichai told Wired, “It’s core to the company to evolve search while applying the underlying principles” – meaning still connecting users to the web. Google even paused its plan to kill third-party cookies in Chrome, partly because their ads team was spooked by the shift to AI (fewer clicks could mean less ad revenue). It’s a tense balancing act: add enough AI to stay modern, but not so much that they break what made Google valuable in the first place.

Which Search Tools Are We Actually Using?

Let’s look at actual usage. Surveys and stats give a fragmented picture. The OrbitMedia poll of 1,040 Americans (just released in May 2025) found 51% of chatbot users plan to use them more, and 49% of respondents think AI will eventually replace search engines. That’s a pretty even split – half are hopeful, half skeptical. They also found that Google and Microsoft (Bing) together now command less than 50% of AI chat tool usage, thanks to newcomers.

To break it down, their data shows about 36% of people use ChatGPT most often, 29% use Google’s Gemini/Bard, 18% use Microsoft’s Copilot/Bing, and smaller percentages use Claude, You.com’s R1, Perplexity, etc. These percentages for chat tools suggest Google is no longer the default platform – at least for chat-based queries. (In 2022, Google was virtually the only player in search. Now it’s one of several in AI Q&A.)

Still, raw search volume paints Google as dominant. Research by SEO experts at SparkToro showed that even if ChatGPT’s entire 1 billion daily messages were “search queries,” Google would be barely touched. In reality, ChatGPT’s daily “search-like” prompts are only about 37.5 million, giving it under 0.25% market share. Microsoft’s Bing Chat has about 4% share, Yahoo 1.35%, DuckDuckGo 0.73% – Google is at a healthy ~93.6%. Plus Google’s raw search traffic keeps growing (up 21.6% in 2024 vs 2023).

In short: casual users are definitely trying out AI chatbots – half of us daily by one survey – and many find them handy for certain tasks (like brainstorming ideas or quick coding help). But as of now, the vast majority of internet searches still start with Google (or another traditional engine) by sheer numbers. It seems we’re in a transition phase: curious minds play with AI answers, yet everyday life still leans on the old reliable blue links.

AI Search: The Good, the Bad, and the “Hallucinating”

What can these AI search engines do better? And where do they still lag? One big advantage: AI chat can summarize complex information. For queries that involve reading multiple sources or fleshing out a how-to, a chatbot saves time. For example, if you ask “How to make sourdough bread step by step?”, a generative AI can pull from baking blogs and recipes and deliver a unified answer. In Google today you might click 3–4 sites and skim them yourself. Many people (myself included) love getting that synthesis.

AI chatbots are also multilingual and multimodal. You can speak or type naturally. Fancy vision models (like Google’s Gemini) can even understand images or videos and answer about them. Soon you might just take a photo of a math problem and get it solved in your chat window.

But there are clear downsides. As one computer scientist quipped, “AI search engines are 100% dreaming and have the hallucination problem; a search engine is 0% dreaming”. In plain English: LLMs sometimes “hallucinate” facts – they make up plausible-sounding answers that aren’t true or not cited properly. They weren’t designed to verify facts; they predict words. For example, people have already spotted AI answers confidently claiming that some person invented a gadget, when in reality that detail was wrong. A recent UW study even showed ChatGPT listing benefits of nicotine that matched smoking addiction research rather than telling the whole truth – the AI literally invented some supporting “facts” that didn’t check out.

Hallucination is a serious concern. Unlike Google, which points you to actual articles, LLMs can just generate an answer. If you don’t double-check the sources, it’s easy to trust something incorrect. (Trust me, I’ve done it. I once asked an AI about the composer of some obscure piece and it cited a non-existent blog!). These mistakes haven’t been catastrophic yet – often the answers are pretty good for general info. But Google’s engineers know this: the Gemini overview warns that AI answers “might be inaccurate, especially about complex or factual topics”.

Another issue: timeliness. Google’s web index is updated constantly; its news and weather answers reflect this morning. Chatbots, on the other hand, are usually trained on data up to some cutoff date (ChatGPT-4, for instance, knows nothing after late 2021 unless explicitly connected to the web). So if I ask a standalone ChatGPT, “Who won the 2024 US election?” it’ll either guess or say “I don’t know.” (Newer tools can browse in real-time, but that’s extra complexity.) Google searches tend to have fresher info by default.

AI chat also struggles with simple navigational queries. The CMSWire article pointed out that if you just type “Facebook login” or “Netflix”, Google quickly gives you the correct site link – which is great if you just want the page. The AI, on the other hand, likes to “think out loud” and may say, “Here’s some info about Facebook…” instead of just linking. That can annoy folks who just want a link. In fact, many AI tools lose out on “this exact site” searches, which are a major chunk of Google usage.

Lastly, there’s the human factor. The MIT study highlighted in TIME found that people who relied on ChatGPT to write essays became less engaged and learned less than those who did their own research (or even just thought alone). In that study, students using Google Search still showed high “brain engagement,” whereas ChatGPT users tuned out. This isn’t exactly a knock on Google specifically – more a warning that easy answers can make us lazy. But it suggests that our habits matter. If we use AI wisely as a tool, great; but if we become too passive and just let AI talk us through everything, we might lose something.

Quotes from the Trenches

It helps to hear directly from the experts who watch this space. Google’s CEO Sundar Pichai is optimistic about the shift. In a recent interview he said, “I think AI is going to be bigger than the internet”. He means that this AI revolution could spawn whole new industries and ways of using tech – think how we can hardly imagine life without the web now. Pichai also insists Google isn’t losing. As Wired reported last year, Google’s internal mantra was loud and clear: “We are not going to lose in AI.” Google is pouring billions into AI research and has reorganized teams (merging DeepMind and Google Brain) to turbocharge its efforts. So from inside Mountain View, the attitude seems to be: “Let’s embrace the change and lead it.”

On the other side, ChatGPT’s chief Sam Altman takes a humbler tack. In a Senate hearing, he joked that Google “didn’t send me a Christmas card,” implying the rivalry is real. But he too believes Google remains strong. Altman quipped “Probably not” when asked if ChatGPT would replace Google. He praised Google as “a ferocious competitor” with a powerful team and a protected business model. In other words, he doesn’t see Google going away.

There are also voices from competitors. Richard Socher, CEO of AI-search startup You.com, is much more bullish on change. He said bluntly: “This sort of insane, untouchable monopoly that Google had for 20 years, those days are over,” and noted that users nowadays switch to new search apps more quickly. If anyone would know, it’s Socher – he used to head AI at Salesforce and saw first-hand how software leaders can be disrupted. (He’s saying: be quick or be extinct.)

And then there’s Elon Musk – cheeky as ever, he retweeted that “AI will obviate search,” referring to his own Grok chatbot. Of course Musk’s tweets can be half-serious, half-provocation, but the sentiment is clear: he thinks AI could make the traditional search experience feel outdated. Whether that happens this year or in five years is anyone’s guess, but it adds fuel to the debate.

SEO, Ads, and the Business of Search

What does all this mean for people who work on the web? If your website relies on Google traffic, you’re right to watch these trends nervously. As we saw, AI Overviews and answer boxes can suck up many clicks that used to go to organic links. SEO (search engine optimization) might need a makeover: instead of targeting keywords, content creators must now add value that stands out even in an AI summary. The consensus among marketers is: “AI search rewards real expertise and depth,” since users and AIs alike prefer quality answers. (No doubt the copywriters and tech bloggers are sharpening their quills now.)

Advertising is also in flux. Google’s entire empire is built on ads shown alongside search results. If users get their answer from an AI snippet, do they even see the ads? Google insists it will find ways to “apply the same principles” – i.e. keep ads in AI modes, just maybe in new slots. How exactly, we’ll see. Maybe the AI will say “Sponsored: Here’s a solution from [Brand]” or give sponsored results a shout-out. Remember how on mobile, Google adjusted to fit ads and quick info together? This is another shift, so expect some learning curve.

One interesting wrinkle: Google has delayed killing third-party cookies partly because advertisers freaked out about AI. With AI, if you’re not clicking around websites anymore, the old ways of tracking users for ads become harder. Google’s Chrome now won’t drop cookies until at least 2025, in part to buy time. The “monetization” of AI search is an open question – will ChatGPT answers someday be ad-free? (Unlikely, someone will have to pay ChatGPT’s bills.) In short: everyone in the ad world is watching closely.

Gazing Ahead: AGI Search and Other Wildcards

All this is just Version 1 of AI search. What’s next, maybe AGI (Artificial General Intelligence) search? It sounds sci-fi, but think of it as a future personal assistant that truly understands you across all contexts. Already companies like Google and OpenAI are talking about AI agents and “personalization”. In the next few years, search might merge with your calendar, email, preferences – the AI will already know who you are and what you like. Instead of searching for “Italian restaurants near me,” you might simply say, “Plan my Friday dinner with Marco at something new,” and the AI books tables and sets reminders.

Google’s vision (just from I/O 2024) is moving in that direction: building custom “apps” on the fly from search, planning trips with follow-up chat, and even video-based troubleshooting. Sundar Pichai imagines devices beyond phones – maybe glasses or ambient computing – where information “surrounds you” and the AI helps manage it. He says it’s an “exciting time to be a consumer”. On the other hand, he cautions it will take a while to move away from looking at that black rectangle (our smartphone screen) to something more intuitive.

True AGI – a machine that can think or learn like a human across any domain – might still be years or decades away (experts disagree). But we can imagine “AGI search” as a scenario where your query triggers a multi-step reasoning across all knowledge. Already tools like Perplexity plug in web browsing (Retrieval-Augmented Generation) to ground answers. As LLM research advances, hallucinations may shrink and reasoning improve. In theory, an AGI could read every relevant document in seconds, compare viewpoints, and even question you for clarifications before answering.

At that point, search might look nothing like today’s SERP. It could feel more like programming a personal digital assistant to do tasks. If you think a smartphone changed how we browse the web, an AGI could change what browsing even means. Sundar Pichai hinted at this shift: “AI is going to be bigger than the internet,” he said. “There are going to be companies, products, and categories created that we aren’t aware of today”. In other words, the whole landscape could be remade.

One cautionary note: privacy and bias become even bigger concerns with AGI. If an AI knows all about me, how is that data used? Google and others are discussing regulations and safeguards right now. I suspect 5-10 years from now, we’ll have had some serious debates (and maybe breakthroughs) on how to keep these AIs aligned with truth and ethics. After all, even today experts like Karpathy remind us “An LLM is 100% dreaming” – meaning it’s fundamentally not grounded in fact unless we make it so. The hope is that as these tools mature, they’ll get much better at saying “I don’t know” or “That might be wrong, check this source.” Sam Altman agrees; he noted that prompting AIs to verify accuracy could become a built-in step.

So, Should Google Be Worried?

In human terms: Google probably sleeps less soundly, but it’s not shaking in its boots… yet. The data tells two stories. On one hand, millions of people (especially younger users) are indeed lured by the convenience and novelty of AI chat. Tech leaders and startups are boldly declaring that the era of Google’s iron grip is fading. On the other hand, Google’s infra and expertise are massive – from web crawling to ads, it’s deeply integrated into the internet’s fabric. Sundar Pichai wouldn’t bet the farm if he weren’t confident; he keeps reminding everyone that Google’s traffic is up and the web is bigger than ever.

As a user and tech pro, my take is this: hybrid is coming. Search will evolve rather than vanish. We’ll likely use AI assistants for certain tasks – quick questions, casual research, or creative brainstorming. But for big decisions, technical issues, or anything where we need to vet sources, humans might still turn to classic search or a mix. If Google can smartly blend AI answers with reliable data (and keep its promise to keep sending traffic to real sites), it will likely remain a cornerstone of the web.

Think of it like this: Google built the library and the card catalog. AI is building a robotic librarian who can even write the summary of a book before you read it. Both are useful. In fact, Google itself calls this a “new phase of the AI platform shift,” not an apocalypse. So I don’t see Google collapsing – but I do see it transforming. It’s as if Google suddenly got a new brain (Gemini), and we’re all on a learning curve figuring out how to talk to it.

In conclusion, should Google worry? Probably a little – competition is fierce, and users’ habits are changing fast. But Google is no stranger to reinvention. They bet early on AI (going “AI-first” back in 2015), and now they’re catching up in consumer AI tools. All eyes are on whether Google’s new search models will really match the slickness of ChatGPT or other friends. As one wise researcher put it, “We should look at the signal and separate it from the noise.” The signal is that AI is a profound platform shift. Google and its rivals will harness it – and we’ll get better search experiences, in one form or another. For now, I’ll keep both my Google and ChatGPT windows open. It’s an exciting race, and luckily, I still have time to enjoy watching it unfold.