marketingtomachines.org

From “Alexa, Set a Timer” to AGI: How AI Changed, and What It Means for Search

A decade ago, talking to a machine meant asking Alexa to set a timer, asking Google Home what the weather was, or telling Siri to play a song. These tools were genuinely useful, but narrow. They followed scripts, matched commands to a fixed set of actions, and got confused the moment a question wandered off the path they were built for.

Today, the same kind of voice or text interaction can mean asking Gemini to plan a multi-step research project, asking ChatGPT to reason through a business decision, or having Claude Code read your codebase, fix a bug, write tests, and open a pull request on its own. That is not an incremental upgrade. It is a different category of tool, and it is changing how people look for information, and how businesses need to be found.

The first wave: helpful, but narrow

When Amazon Echo launched in 2014 and Google Home followed in 2016, they kicked off the smart speaker era. By the early 2020s, smart speaker ownership in US households climbed from about 7% in 2017 to roughly 35% by 2022, and Amazon Alexa alone now runs on more than 600 million devices worldwide. Voice assistants overall reached around 157 million users in the US.

But look at what people actually used them for. The most common Alexa requests were setting timers and alarms, playing music, and getting news updates — the kind of bounded, single-answer tasks these systems were designed for. Even at their best, accuracy on general questions hovered around 80%. These were assistants in the literal sense: helpful for small, well-defined jobs, but not something you would trust with an open-ended question or a multi-step task.

The leap: from assistants to something closer to AGI

The shift since the launch of ChatGPT in late 2022 has been staggering, both in capability and in adoption.

On capability, today’s leading models do things that would have sounded like science fiction in the smart speaker era: holding extended reasoning conversations, analyzing documents and images, writing and debugging software, and increasingly, taking actions on a user’s behalf rather than just answering questions. Claude Code is a good example of this shift. It is not a chatbot that tells you how to fix code — it is a tool that can read a project, make the changes, run the tests, and report back, the kind of agentic behavior that marks a real step toward general-purpose AI.

On adoption, the numbers reflect that shift:

For comparison, it took smart speakers roughly six years to go from 7% to 35% household penetration. Generative AI tools crossed from niche to mainstream in a fraction of that time.

So is search dead? The data says: not yet, not close

Here is where it gets interesting, because despite all of that growth, traditional search has not collapsed. If anything, the data shows just how durable it still is.

Google still holds about 90% of global search market share and around 85% of the US market, processing an estimated 13.7 billion searches per day — more than 5 trillion per year.

For every search-like query on ChatGPT, there are still roughly 8.33 equivalent searches on Google. Even with ChatGPT’s explosive growth, its search-like query volume is estimated at around 12% of Google’s.

64% of US teens have used an AI chatbot, and searching for information is the most common use case, according to Pew Research — but that has not pulled them away from Google entirely. Research from Semrush found that after people adopt ChatGPT, their Google usage does not decline; people simply use both tools for different purposes.

Search results still directly shape buying decisions. Google reports that 39% of purchasers made their choice as a result of a relevant search, and a large share of consumers still favor organic search results over ads when deciding where to click.

In other words, for the vast majority of people, on a normal day, typing a question into a search bar is still the default way to look for information. That has not changed nearly as fast as the headlines about AI might suggest.

But the lines are blurring fast, and that is the real story

Here is the pivot. While search has not been replaced, it has started to behave less like “search” and more like an answer engine — and that shift is accelerating quickly.

Google’s AI Overviews appeared on about 48% of tracked queries in February 2026, up from 31% just a year earlier — a 17-point jump in twelve months, according to BrightEdge.

Zero-click behavior is now the norm, not the exception. Roughly 58.5% of US searches and 59.7% of EU searches end without a single click to an outside website, as answers, summaries, and AI-generated overviews satisfy the question directly inside the results page, per Semrush.

While total search volume keeps growing in absolute terms, per-user query volume on Google is down an estimated 20% year over year — what some analysts are calling “the Great Decoupling”: the overall pie keeps growing, but the slice that used to lead to a website click is shrinking before it ever leaves the search results page.

37% of consumers now say they start their search with an AI tool rather than Google — a dramatic shift from just two years ago, even though Google still wins on raw volume.

Zoom out further and the picture gets clearer still. One analysis found that Google’s share of all “information discovery” — search plus AI assistants combined — fell from about 89.3% in December 2022 to roughly 57.6% by December 2025, even as the total discovery landscape grew by about 26% over the same period. People are not searching less. They are spreading that search behavior across more tools, and increasingly expecting an answer, not just a list of links.

What this means going forward

Put it all together and the trajectory is clear. Search is not disappearing — it is the foundation that AI systems are increasingly built on top of, and the interface layer is shifting from “ten blue links” to direct answers, summaries, and increasingly, autonomous agents that complete tasks rather than return results.

As search engines themselves get more sophisticated — integrating AI Overviews more deeply, rolling out agentic shopping and research features, and blending traditional ranking with generative synthesis — the gap between “search” and “AI” will keep narrowing. The businesses that show up well in this new landscape will be the ones whose content is structured not just for human readers clicking through a list of results, but for AI systems summarizing, citing, and recommending them directly. That is the shift this entire site is built around: marketing in a world where the audience reading your content is increasingly a machine, acting on behalf of a person.

Sources and further reading