Should we be worried about Google Zero?

In 2024, Nilay Patel the CEO of The Verge coined the term Google Zero. He was warning of the day when Google decided to switch their default from search to AI, and how it could threaten to drive most websites’ publishers to zero. 

That day is here. Or at least it’s been announced. 

At the recently concluded Google I/O conference, one of the big news pieces was that search as we know it, which would return us with links, would be replaced. In its stead we will be given AI-powered and sometimes interactive experiences. In an interview with Patel on the Vergecast podcast, Google CEO Sundar Pichai gave this example: if before you searched for black holes, you would be given links to websites you could click into and read, now when you ask google about black holes it might return not just with text, but agentic AI might build you an interactive website that allows you to learn about black holes. 

RELATED STORY: Google highlights Gen Z creators at Search Day

On its face this actually sounds cool. But beneath it are a number of questions worth asking. Of course at this point we are mostly anticipating what this is going to look like and how this will affect things. Still, given the speed of change and the tremendous potential impact it’s important that we start asking now so we can make decisions for the future. 

Publishers hit first

If people “search” for something and instead of getting links they get an AI overview, then most of the time they stop there. A study from last year showed that if an AI overview was provided, only 8% of users would click into the search results or provided links.  

So basically, if someone searches and Google provides them with an answer, then most users have no inclination to check the accuracy of that summary, to click through the sources to confirm, and to find out more. This is going to have a much larger effect on how we process information, but I’ll address that in a later section. 

First, I’ll look at this from the “publisher” perspective, or basically from the perspective of anyone who puts out content like articles on their website to generate visits and clicks. What Google has said is that while overall search traffic will go down, the quality of click throughs will increase, as those who are visiting your website are going to do so more intentionally. The numbers don’t support this, but we will have to wait and see how this turns out overall. 

The real challenge will be for publishers (like news sites) whose key metrics include page views and click-throughs on ads. Publishers have relied on click-throughs and ad placements since these have been available. But lower page traffic will mean lower placements and lower click-throughs, which means lower revenue. 

Granted the value of this has been plummeting in the last few years, but this Google Zero event really challenges publishers to ask, what happens if instead of people showing up thanks to Google search results, Google just kept these users to themselves. By building these “AI-powered experiences” Google disincentivizes users to visit the actual websites where information is housed, choking those sites of visits and by extension possible revenue. As if to add insult to injury, those experiences and results will still be based on the content of those available websites. 

You can expect that thanks to Google Zero many publishers who you rely on for information will start to explore other revenue models. If they can’t make money off page visits and advertising, then they’ll need to explore things like paywalls, subscription models, and other alternatives. 

Now note, this isn’t the first time that Big Tech has made a move that dealt a crushing blow to publisher revenues. Not that long ago social media was a driver, when publishers were encouraged to post on social media networks and then to put links, and those links led to lots of page visits. That was until social media companies decided that they didn’t like users leaving their platforms and going out to other websites, so they throttled posts with links that led people to click out. 

This is Google just taking their own shot, saying they would prefer people stay inside of Google than visit other websites. They would rather burn tokens and “build experiences” rather than encourage people to click away. I don’t know if this triggers a necessarily adversarial response from publishers, but my read is this is a decidedly aggressive move on the part of Google to take away publishers’ lunch. 

RELATED STORY: Everything launched at Google I/O 2026

While brainstorming the pitch for this piece, TechINQ editor Jayvee Fernandez asked me, “Will traffic still matter?” And at this point, I think that we will start seeing traffic very differently. If social and search are already defaulting to hold onto users and deprive publishers of traffic, then metrics and tactics will change. So it’s not that traffic won’t matter, but soon we will both define it and then measure it differently. 

What this could mean for the internet in general

When publishers can’t get traffic from search and their ad revenues will be affected as a result, then they will be forced to explore other means of generating revenue. In addition, if you leave your content open and free to access, then Google will use whatever you’ve posted in creating its “interactive experiences.” There is no indication that there will be some kind of compensation or payment model for when your site gets cited or used in creating any of these search results. 

These two factors: the inability to generate revenue and the use of content without compensation for “query results” are likely going to affect the availability of good information on the internet. 

Good information, like journalistic reporting, thoughtfully written essays, creative writing, fun content, and all of these things that we actually enjoy from the internet take time and money. However, misinformation and disinformation, AI slop, Slopaganda, and generally divisive content are pretty easy to generate and publish on the internet. 

Put simply, good stuff costs more money to make, and bad stuff can be produced cheaply. And unfortunately as we have learned from the many studies on information operations, there’s plenty of money being used to produce what can be called information disorders. 

What does this have to do with Google Zero? As publishers try to find revenue models that will sustain their more expensive operations, we are more likely to see paywalls and other means of monetization. The last 15-20 years have unfortunately conditioned us to the idea that news and information are free. But what’s true is that publishers will need to find revenue and asking us to pay for access is the likely outcome. 

But this means that the integrity of openly available information will go down. And then what’s free and openly available will either be publicly or philanthropically funded (best case scenario) or sustained by less than reputable means. 

Just as an example, think of how much disinformation noise is on your social media feeds vs. actual good stuff. Now project this out into AI search results, where publishers will hide good content from their AI search agents, which means that those agents will wind up drawing from openly “free” sources that don’t have paywalls and AI-blockers. 

Google search was already pretty bad because of how you would often get a page of sponsored and not good results– that was a design decision made by Google. Now you’ll get some really good “experiences” but you won’t be able to easily discern whether that information is reliable because the internet landscape will likely change for the worse. 

What this could mean for us as individuals using search

I am old enough to remember when “GMG” or “Google mo, g@g0” became a thing. What it really meant at the time was the information was out there, and you didn’t take the time to even bother checking Google for the right answer. But this speaks to how meaningful and important Google search was in culture: it had become a verb, it was the brand-name stand-in for internet search, and it indicated a kind of trust in its use where if you could just google something, you could find the answer. 

As mentioned earlier, google search as a product has gotten what Cory Doctorow calls “enshittified” in that it doesn’t return results in a way that serves the user. But in this next phase, it removes even the agency of searching through the provided links. It will provide us with answers and agentic or AI-enabled experiences. 

Why is this possibly a bad thing? Because by providing us with answers instead of links, it positions google as the start and end point for information collection. It says, hey, don’t worry about trying to figure this all out by yourself, we will give the answer to you. 

You might ask, how is this any different from something like Wikipedia providing us with answers, or the olden days of looking for information in an encyclopedia. There’s two layers here: the kinds of information we will get, and then how this will condition us to think. 

First is the kind and quality of information. I am also old enough to remember the moral panic around Wikipedia and how people would stop reading and learning because everything was just there. But as we have seen, Wikipedia feels like a living database of information, regularly updated and curated. We also still treat it as a source of starting point research but are all very aware that there are editorial processes around it, as well as the fact that there can be attempts to put bad information there. With old encyclopedias there were editorial and process standards in assembling those volumes, people doing the fact-checking work for the entries, and so on. 

Now if we look at what Google is proposing to give us now, it’s basically, LLMs providing us with the information. And as we should have learned in the last few years, hallucinations will always be a part of LLMs. That’s just part of the design. If even now Google’s AI overviews can provide us with wrong or incorrect information, then one can only hope that these get significantly better. 

As I identified in the previous section, if publishers who are making factual and accurate content are more incentivized to hide that content behind paywalls and AI-blockers, then whatever these answers generate will be limited to training data and what’s still available on the internet. I think that sure, if you’re researching something about black holes, which is a very technical topic with a lot of papers and materials about it, then you could probably get a reliable answer. But what if it isn’t something with as much ground truth and reliable literature? What if it’s new trends or ideas? Or important topics that might not have as much varied coverage? 

Another thing to add on here is the question of what models and how many tokens will be devoted to each query? If you’re already using chatbots, you know that cheap and fast models can’t give you answers as good as the more powerful models. But using those more powerful models consume more tokens, and if you’re a power user then you’re probably maxing out your token allotments. 

When search becomes AI-powered answers, what models will be in use? Will we get super cheap and fast models? Between the prospect of that and a possible dearth of reliable open sources to build answers, that’s a problem. If instead Google gives us powerful models for search, then that means using a lot of token processing. Imagine if your search for black holes creates a whole app for you to interact with, how much computational power does that consume? And eventually, who pays for that? 

Now we get to the last question, which is what is this going to do to the way that we think? If we know that people have already started turning to google search results in AI mode for truth, then that means that they aren’t checking sources, they aren’t worried about hallucinations, and that it’s likely that objective truth really is going to be a problem. As a famous example, if we followed what the LLM told us, we would all be putting glue on our pizzas. 

As with that example, because we are humans with experience of the world, we have enough sense to know not to put glue on our pizzas. But what other things could an LLM recommend to someone who was using Google to get answers? And how much discernment would people have each time, especially with regard to topics that they don’t have much knowledge of? 

I sound like an advocate for search links. But what I’m actually advocating for is people go through a process to get answers, especially with complex and complicated topics. Knowledge acquisition should be a process, with looking at sources, discerning their veracity, and then coming to our own understanding through thought and analysis. What this new version of Google could possibly do, whether they intend it or not, is encourage people to just take whatever answer it spits out. Think especially if the information is faulty, but it is wrapped inside of a very well-constructed learning experience. 

This is a reaction to Google Zero, but it’s also my invitation to everyone to start thinking about how we are interacting with the internet that is being built around us. The past few weeks the term “open web” has been used a lot, and often there have been warnings that the time of the open web is over. The walls and gates are coming up. The end of Google search as we know it, on the interface, is going to be a small, barely noticeable change when we open up the app or website. But it will have real ramifications for how we access and process information.