Forget Google Searches: The 7 AI Habits Quietly Rewiring How We Find Things

A viral round-up of “7 AI prompts” to replace Google is really a symptom of something bigger: the search box itself is splitting into two separate tools. Here’s what that split means for anyone in Singapore or Southeast Asia building a business, a brand, or a website around being found.

Somewhere in the last few months, “just Google it” quietly stopped being the automatic reflex it was for two decades. A wave of articles telling readers to forget Google searches and lean on seven AI prompts instead has been making the rounds, and on the surface it reads like just another productivity listicle. It isn’t. It’s the visible tip of a structural change in how information gets discovered, and the numbers underneath it are more dramatic than the headline suggests.

The story has already been covered from dozens of angles: how-to guides on disabling Google’s AI Overviews, roundups of clever prompts, and breathless “search is dead” proclamations timed to Google’s own product announcements. What’s missing from most of that coverage is the part that actually matters to anyone who runs a website, sells a product online, or simply wants their business to be findable: this isn’t a story about one search engine losing to another. It’s a story about two different habits of finding information developing side by side, each with its own economics, and most businesses are only prepared for one of them.

What’s actually new here

Search behaviour has fragmented faster than most marketing teams have adjusted for. Roughly 37% of consumers now say they start their searches with an AI tool rather than Google, according to data compiled by Search Engine Land. ChatGPT’s weekly user base has more than doubled in about eighteen months, climbing past 900 million from roughly 400 million in early 2025, which by some measures makes it the fourth most-visited destination on the internet.

Google, for its part, hasn’t stood still. At its most recent developer conference, the company unveiled what it called the biggest change to the search box in more than 25 years: an “intelligent search box” that can drop users into interactive AI experiences, dispatch autonomous “information agents” to research on their behalf, and let people build small personalised tools instead of just returning a page of links. Google’s own AI Overviews now reach roughly 2 billion monthly users across more than 200 countries, and its standalone Gemini app has grown to around 750 million monthly active users.

So the popular framing, that people are abandoning Google for ChatGPT, undersells what’s happening. Google is racing to become an AI answer engine too. The real shift is that the open web’s front door, a page of ranked blue links that anyone could click through and compare, is being replaced on both sides by a single synthesised answer. That has consequences most coverage of this trend hasn’t connected yet.

The backlash is the more interesting half of the story

What tends to get left out of the “AI is replacing search” narrative is that a meaningful slice of users are actively fighting it. Browser extensions built specifically to strip AI Overviews out of Google results have accumulated well over 100,000 installs. Community-run, self-hosted search front ends built to route around AI summaries have seen usage spike. DuckDuckGo has held its line as one of the only mainstream search engines that still returns plain organic links by default, no AI overlay required.

The tension isn’t “humans versus AI search.” It’s two camps of the same audience: people who want an answer synthesised for them in seconds, and people who still want to see the sources and decide for themselves. Most businesses are only building for the first camp.

Why this matters more if you run a website in Singapore or Southeast Asia

For a regional content publisher, this split isn’t an abstract trend piece, it’s a traffic and revenue question that arrives faster than most people expect. Singapore and its neighbours have some of the highest smartphone penetration and mobile-first browsing rates in the world, which historically made them fast adopters of whatever interface search engines pushed next, from voice search to visual search. There’s no reason to expect AI-first search to be an exception.

Consider the mechanics. When a search engine answers a question directly inside an AI summary instead of sending the user to a website, that website loses the click, even if its content was the source the AI drew from. Publishers in the US and UK have already reported measurable referral-traffic declines tied to the rollout of AI Overviews. A regional or niche site, the kind covering cycling gear, golf news, or a specific product category, is arguably more exposed than a major news brand, because it has fewer alternative distribution channels (a loyal newsletter list, a strong social following, a branded app) to fall back on when search referrals soften.

That’s the local angle wire coverage of this story keeps skipping: the sites most likely to feel this shift first are not the big publishers being quoted in the trend pieces, but small and mid-sized regional sites whose entire traffic model was built on ranking for search queries that an AI answer box can now resolve without a click.

A useful historical comparison

This isn’t the first time the front door to the internet has been redrawn. Portal-era homepages like Yahoo and MSN once decided what most people saw online, until Google’s clean, link-first search page made portals feel bloated and slow, and traffic quietly migrated away from them over a few short years. Site owners who adapted, by building direct audiences instead of depending entirely on portal placement, survived that transition; those who didn’t largely disappeared. The AI-answer shift looks structurally similar: a new interface layer is inserting itself between the user’s question and the open web, and the businesses treating it as a temporary annoyance rather than a permanent structural change are the ones most likely to be caught flat-footed a second time.

Seven habits, one underlying pattern

Strip away the specific prompt examples from these “forget Google, use AI instead” guides and a consistent pattern emerges across all of them. People are using AI tools for the same handful of jobs, over and over:

• Turning a vague need into a personalised, multi-step plan (an itinerary, a routine, a shopping shortlist) rather than a list of links to sort through themselves
• Asking for a synthesised comparison between two or three options instead of opening ten tabs
• Getting a running explanation that adapts as they ask follow-up questions, rather than starting a new search each time
• Using natural, conversational phrasing instead of the clipped keyword strings that used to work best on Google
• Treating the AI’s first answer as a starting draft to refine, not a final verdict to accept
• Reaching for AI specifically to avoid ad-heavy, SEO-optimised pages that dominate traditional results
• Still circling back to traditional search, or a direct site visit, for anything transactional: buying, booking, or verifying a fact that carries real consequences

That last habit is the one most “Google is dead” commentary quietly ignores. People trust AI to plan and explain. They’re far more cautious about trusting it, alone, to transact or to fact-check something important, which is exactly why classic search hasn’t disappeared even as its share of first-touch queries falls.

What conflicting coverage reveals

Read enough of the surrounding coverage and a genuine disagreement surfaces. Some outlets frame Google’s AI Overview rollout as the company successfully co-opting the AI trend before it could be disrupted. Others frame the same rollout as proof that Google is under siege, pointing to publisher traffic complaints and a fast-growing cottage industry of tools built purely to switch AI features off. Both readings can be true at once: Google’s usage numbers are climbing and its market position, measured by total queries, remains dominant, while the quality of that traffic, measured by whether anyone actually clicks through to a website afterward, is deteriorating for the publishers who depend on it. The headline metric and the metric that pays the bills have quietly come apart.

The takeaway