It started out as a rather simple function. It’s strange how much search has changed, and how it is changing us.
There is probably no digital record more intimate than a search history. Our social media is carefully curated. Even private messages are written for someone else. Search may be one of the few non-performative things we still do online: a record of questions asked without an audience, often before we have decided whether a thought is worth saying aloud.
British data journalist Simon Rogers’s new book, What We Ask Google, begins with the proposition that this peculiar archive can tell us something about ourselves.
Drawing on two decades of Google data, he treats this function as a social mirror, and what he finds is more humane than one might expect. People want to know how to help a bee, how often they can donate plasma, how to support someone who is depressed or grieving. In times of crisis, they look for ways to give blood, donate money and help strangers.
Googling doesn’t constitute action, and curiosity is not really a virtue. But attention and intent are not nothing either, particularly when expressed privately.
Google says it handles more than five trillion searches a year, and roughly 15% of these have never been made before. That works out to more than two billion unfamiliar queries a day. Not necessarily two billion new ideas: a different combination of words, an obscure personal predicament, or a question about something that happened five minutes ago can all count as new.
Still, that is quite remarkable. Nearly three decades in, humanity apparently remains capable of surprising the engine. Our curiosity has proved almost impossible to exhaust.
A QUIET SHIFT
Now we get to some interesting twists.
Rogers’s book captures what might be called the Search Era, when the essential bargain was straightforward: we formulated a question and the machine helped us find the sources from which an answer might be assembled.
In aggregate, our questions became an extraordinary record of what people wanted, feared, desired, misunderstood and cared about. The data reveals sleepless parents and anxious patients, heartbroken lovers as well as people trying to figure out how to boil an egg.
The profound and banal coexist without embarrassment.
This, of course, is not a census of human curiosity. Mainstream search is shaped, more than anything, by immediate need. We rarely ask the deeper questions of this task bar.
The results are also moderated, and shaped by laws too. Darker forms of inquiry migrate elsewhere. Rogers’s mirror, therefore, is large but bounded.
What’s truly interesting about Google’s records is how it lays bare the ways in which the search engine has shaped our habits.
IN THE DEEP END
About 15 years ago, American journalist Nicholas Carr argued, in The Shallows (2010), that the internet was not simply giving us more information; it was altering how we viewed the idea of information itself.
The web rewarded scanning, switching and fragmentation. Retrieval became so easy that remembering seemed less necessary. (Why retain a fact when it can be looked up in seconds?)
There is no reason to romanticise the inconvenience that preceded this. Humans have always outsourced cognition. Writing externalised memory, books preserved knowledge, maps externalised geography, and calculators relieved us of arithmetic.
Search is part of that tradition. The difference lies in its reach. It placed an almost unimaginably large external memory within instant reach, all the time: Text, photos, videos, eventually footage from space and live satellite feeds.
Take Google Earth. For most of history, seeing a distant landscape required us to travel there, access a photograph or rely on a description. Now, we can “walk” through a city, stop on a street and find a specific front door, all on another continent, without moving from the couch. We get the access, but little of the history, memory or lived experience of being in that place.
Which poses the question: Can access create its own illusion of understanding?
TOO MUCH?
This has been the deeper transformation of the Search Era.
The early internet presented a problem of scarcity: how to find something in an expanding wilderness of information. Its mature form presented a problem of abundance: how to decide what among millions of possible results deserves attention, retention and thought.
Meanwhile, as millions of results competed for attention, their ordering became a form of power. What looked like a list of answers was also a marketplace of advertisers, optimised content, affiliate recommendations and businesses that paid for the privilege of appearing first.
Over time, we adapted to this arrangement. We learned to trust differently.
The ranking of answers, which was never a neutral representation of knowledge, took on a sense of authority.
We didn’t just stop questioning that. We stopped wondering what had been excluded, or why. We grew accustomed to the idea that an algorithm could determine what parts of the world were visible to us.
DIVE BELOW
Artificial intelligence (AI) represents the next twist in the road.
We are moving from the Search Era to the Answer Era.
The English poet John Dryden wrote, more than three centuries ago: “Errors, like straws, upon the surface flow; He who would search for pearls must dive below.” Google made the dive easier. It could take us rapidly to the right part of the ocean. But we still had to open the links, read the sources, compare them and decide what to believe.
AI increasingly offers to make the dive for us.
According to Google, as of May this year, AI Overviews reach more than 2.5 billion people a month. The Answer Era is already operating at scale.
What does this mean for how we think?
Carr worried that the web was producing cognitive shallows.
The AI age may produce something different: epistemic autopilot, which I define as a condition in which we delegate more of the navigation, synthesis and judgment to the machine.
It comes with a cultural companion: answerism, or the expectation that uncertainty can swiftly and reasonably be resolved.
What is lost in the process can feel hard to articulate. Difficulty is not intrinsically valuable, and much of the friction eliminated by technology deserves to disappear. Nobody becomes wiser by spending an afternoon locating a fact that could take 10 seconds to find.
But some friction has value. Reading several accounts, discovering that authorities disagree, and deciding what evidence deserves trust are not obstacles placed in the path of understanding; they are often the substance of understanding itself.
NET LOSS
Instead, what the two-second search has given us is a restlessness with details; an impatience with nuance.
I think of this as epistemic regression: a growing discomfort with the unfinished nature of knowledge.
We ignore the fact that reality does not resolve itself into satisfactory summaries.
A POST-QUERY WORLD
Even the Answer Era may prove transitional.
Because another element of traditional search is beginning to disappear: the query.
We are at the beginning of what might be called the Post-Query Era.
For most of the history of the internet, we had to translate desire into language. Even an inelegant Google query required us to name the thing we wanted. Visual search altered that relationship first. With Google Lens, the object itself can serve as the query: point a camera at a plant, a shoe, an unfamiliar object and answers appear.
The path to ambient search is already taking shape.
An audio assistant places an answer directly in the ear. Smart glasses can make visual information available without the need to take one’s phone from one’s pocket.
Further along this path, brain-computer interfaces suggest a future in which the distance between requesting information and accessing it becomes smaller still. The defining feature of the Post-Query Era could be the end of searching as a distinct act.
The context itself becomes the query.
An updated game score or election result is delivered to us because the system knows what game we have been following. Already, coffee and toilet rolls arrive at the door because we logged our usage history and turned on auto-order on Amazon.
When we don’t need to formulate the question at all, who is it that really controls our view of the world?
BUY THE WAY
It is also worth considering that convenience depends on intimacy. A system can understand (or predict) a request because it possesses context: what we searched, read, watched, bought, photographed, where we went and what we ignored.
Personalisation and surveillance are often the same technical capability, viewed from different angles.
Commerce melded with AI adds a further complication.
Sponsored results have so far been comparatively visible. In an AI-mediated environment, the boundaries require greater scrutiny. If a system recommends a hotel, an insurance policy or a new phone, does it reflect our past behaviour? Or is it really working for someone else?
When it recommends an article or prioritises certain data, is it echoing what we already think (which is bad enough), or working to a different brief entirely?
If a sponsored link becomes a sponsored chain, in a world of algorithmic order, is it still search at all? Or a walled garden covered in billboards, blocking out the rest of the landscape?
HOLD ON TIGHT
For now, there is something worth cherishing in what Rogers’s catalogue reveals.
We encounter something we do not understand, or something we care about, and decide that it deserves a question. Sometimes the question is trivial or selfish; surprisingly often, it concerns somebody else: how to help, explain, or care.
Our curiosity is still ours; it is edged with concern for others.
That feels like something worth holding on to.
(Kashyap Kompella is a tech-industry analyst and author of three books on AI)
