Google is changing: Are we ready?

Photo by sarah b on Unsplash

Do you remember a time when the word ‘Google’ wasn’t a verb? When we didn’t casually ‘Google’ any question that came to mind or problem that needed to be solved? 

Since its launch in 1998, Google has become a fixture in our lives. And almost two weeks’ ago, at the Google I/O conference, it was announced that the search bar, which we have used for 28 years as a way to access sources that advised us on everything from how to cook a roast dinner to why we still do not have peace in the Middle East is undergoing the greatest change in its existence.

We are already familiar with the AI overviews that crowd our screen and tempt us to stop scrolling further to reach the actual sources when we now seek information. Soon, we will be engaging with an intelligent search box that will encourage us to converse with Google, to initiate agents to continue our searches behind the scenes and possibly, to never have to go to a secondary source (Perez, 2026). 

Replacing the tedious link-hunt with instant access to complete, cohesive answers sounds fantastic. But the transformation of Google from a search engine to an answer machine has far greater implications than simply saving time. 

The Structural Implications of AI Search Engines

The transformation of Google from a search engine to an answer machine is not merely a product update — it is a fundamental restructuring of how information flows between those who create knowledge and those who consume it. Understanding what is at stake requires examining the shift across three interconnected dimensions: cognitive, economic, and social.

The biological toll: what frictionless answers do to the brain

We may bypass the processes that actually build knowledge. The shift from active, effortful search to consuming AI-synthesised answers allows users to completely skip the generative work essential for deep learning (Klein and Klein, 2025).

Schemata get short-circuited. The brain encodes new information by connecting it to existing internal structures called schemata. When we offload all information processing to AI we default to fast, intuitive thinking (what Kahneman calls System 1) rather than engaging the slower, analytical System 2. Oakley et al. (2025) argue that excessive cognitive offloading produces shallow schemata: weak mental frameworks that cannot support critical reasoning or advanced problem-solving.

Critical thinking circuits atrophy. Klein and Klein (2025) introduce the concept of the “hollowed mind” to describe this risk: a cognitive state where users chronically bypass deep processing, mistaking access to information for genuine ability. Their paper draws on neurobiological evidence to show that the prefrontal cortex networks responsible for critical thinking are strengthened through conflict detection and error resolution, and weakened when AI systems consistently serve smooth, plausible answers that remove the need for those cognitive loops. As they put it, “the very cognitive effort that AI promises to relieve us of may be precisely the effort our brains need in order to learn” (Klein & Klein, 2025, p. 4).

The economic collapse of the open web and the rise of the slop economy

The cognitive risks of AI search do not exist in isolation. They are bound up with a structural collapse in how information gets produced and funded online.

The traditional internet is running on an unspoken agreement that AI search is breaking. For thirty years, websites allowed search engines to scrape their content, and in return, those engines sent traffic back to the creators. As BBC technology reporter Thomas Germain explains, “Now, there’s a new exchange. Google scrapes your information free, and you get maybe nothing at all” (Germain, 2026). Zero-click searches, where users begin and end on Google without visiting any source, have already climbed to around 60% (Germain, 2026).

The result is a bifurcated web. Read (2024) describes the “slop economy”: a thriving, global grey market of AI content mills and algorithmic spam that floods search results with content of  “limited factual nutrition.” As approximately 40% of content consumed online is already AI-generated (Spennemann, 2025), high-quality, human-curated publications will retreat behind premium paywalls to earn the money previously accrued through web traffic. 

A new digital divide

The original digital divide focused on who had physical access to the internet. That gap is closing: 5.5 billion people are online as of early 2025, and by 2030 nearly 90% of people aged six and over will be connected (Statista, 2025, as cited in Miklian & Hoelscher, 2026). The emerging divide is qualitatively different.

The new divide splits along information quality, not access. Miklian and Hoelscher (2026) describe two populations: “digital elites,” who have the financial means and skills to access premium, fact-checked sources, and “digital commoners,” whose online experience is dominated by the slop economy. The divide is largely determined by an individual’s ability to pay their way past the paywalls that will ring-fence quality content.

AI functions as a leveller for novices but an amplifier for experts. This is what Klein and Klein (2025) call the “Expertise Duality.” For novices who lack internal schemas, AI functions as an external scaffold that completes tasks without building the underlying cognitive structures needed to evaluate or improve those outputs. For experts with rich internal knowledge, AI genuinely amplifies capability, because they can validate, interrogate, and improve what the AI produces.

Accelerated silos and the spread of misinformation

One of the subtler structural risks in this shift is the gradual erasure of source awareness.

When users receive direct answers rather than source lists, they lose sight of where knowledge comes from. Germain (2026) describes this as an enormous philosophical problem: it used to be that evaluating where information came from was a central part of the search process. That work has now largely been removed from the user experience. This fosters a dangerous dependency where a single monolithic technology company is viewed as the singular voice of objective reality, rather than a mere gateway through which original knowledge is navigated.

Large language models are prediction engines, not verified databases. Germain (2026) reports directly from his own experiment: by publishing a single satirical blog post, he was able to make both Google’s AI search and ChatGPT present fabricated information as fact within 24 hours. The system was easily manipulated, and the misinformation persisted for weeks. “This is a manipulation tactic that is being carried out on a massive scale and affecting the information that billions of people are seeing every day,” Germain observes (2026).

What this means for how we learn, work, and decide

The picture that emerges from this research is not simply about a search engine upgrade. Three forces are converging: AI search is stripping away the friction that actually builds knowledge; the open web is fragmenting into a slop economy that rewards volume over truth; and a new digital divide is hardening — not along lines of access, but along lines of competence. Together, they are creating conditions where the ability to think critically about information is simultaneously more essential and less commonly developed than at any point in the internet age.

The response to this cannot be to reject AI tools. It must be to ensure that the people using them have the internal knowledge structures, evaluative habits, and source-awareness to use them well. Klein and Klein’s expertise duality makes this point sharply: AI amplifies those who already know enough to interrogate it. Building that knowledge base — in students, in teams, in organisations — is not a supplementary goal. It is the precondition for everything else.

This is the work of information literacy, and it has never mattered more. At Linking Learning Advisory, I partner with educational institutions and organisations to design learning experiences that develop exactly these capacities: the critical thinking, digital discernment, and evaluative skills that AI cannot replicate and cannot replace. If the gap between digital elites and digital commoners is largely a gap in competence, then the most strategic investment you can make — for your students, your workforce, your organisation — is in the thinking skills they bring to the tools they use.

Google has changed. The question is whether we are ready.

References