Summary

Artificial intelligence is changing how people navigate the internet. Instead of searching, opening websites, comparing sources, and assembling an answer ourselves, we can increasingly ask a question and receive a synthesized response. For a large amount of everyday information, this is simply more convenient.

History suggests that technologies built around convenience rarely produce a simple transition from an old behaviour to a new one. Industrial food changed how people ate, but eventually contributed to a wider interest in ingredients, nutrition, provenance, and cooking. Social media delivered on much of its promise to connect people and democratize communication, while also producing concerns around mental health, algorithmic influence, polarization, and the quality of attention. Neither technology was rejected. Instead, people began searching for a better balance between what it gave them and what it displaced.

AI may eventually produce a similar response. As it becomes better at answering, summarizing, recommending, and comparing, it could reduce the need to visit many of the websites that currently sit between a question and an answer. But people do not use the internet only to retrieve information. They also browse, follow unexpected paths, compare perspectives, discover unfamiliar things, and sometimes develop their understanding by looking around.

If those behaviours remain important, the future may contain both an internet that comes to us through AI and an internet we deliberately choose to explore. That could create a different role for websites, with more value placed on becoming destinations people want to enter rather than containers of information that can be summarized elsewhere.

Convenience rarely ends the story

The history of convenience technologies suggests that adoption rarely represents the end of the story. New systems become popular because they solve real problems, but their wider effects are often harder to see until they have been absorbed into everyday life. What follows is not always a rejection of the new system. More often, people begin deciding which parts of it they want to keep and which parts require a counterweight.

Food offers a useful example. Industrial processing made food safer, more durable, easier to transport, easier to prepare, and available at greater scale. Those benefits mattered as patterns of work, family life, retail, and urbanization changed, and convenience food became part of the infrastructure of modern life rather than a temporary cultural experiment.

By 2012, more than three quarters of the calories contained in packaged food and beverages purchased by U.S. households came from moderately or highly processed products. Ready-to-eat and ready-to-heat products accounted for most of the energy in those purchases. [1] (PubMed)

Over time, another food culture developed alongside that system. Consumers became more interested in ingredients, nutrition, processing, provenance, organic agriculture, local food, and cooking. Ideas that once belonged to specialist health movements gradually moved into supermarkets, restaurants, advertising, and mainstream consumer culture. USDA data shows U.S. organic food sales reaching an estimated $65.4 billion in 2024, after decades of growth in consumer demand. [2] (USDA)

That did not mean convenience disappeared. The International Food Information Council’s 2026 Food & Health Survey found that convenience had overtaken healthfulness as a driver of Americans’ food and beverage choices for the first time in the survey’s 21-year history. [3] (IFIC)

The more useful lesson is that people can continue embracing convenience while developing stronger expectations around quality, ingredients, source, preparation, and health. A system can become deeply embedded in everyday life while still producing a desire to recover some of what became less visible along the way.

Social media followed a similar pattern

The early internet carried its own expansive promise. It could connect people across geography, expand access to information, create communities around shared interests, and allow individuals to communicate without relying entirely on traditional institutions. A 2006 Pew Research Center study found that internet use helped people maintain geographically dispersed social networks and seek information and support beyond their immediate physical communities. Rather than simply replacing offline relationships, online communication often supplemented them. [4] (Pew Research Center)

Social media extended that possibility. Anyone could publish, friends could stay connected across continents, people with niche interests could find one another, and creators could build audiences without traditional gatekeepers. Communities that were difficult to sustain locally could exist globally, and much of that original promise remains part of what makes social media valuable today.

What changed was the environment surrounding those possibilities. Platforms became increasingly sophisticated at selecting what people saw. Feeds learned from behaviour, while recommendation systems determined which posts, videos, opinions, advertisements, and creators appeared next. The experience gradually became less dependent on people deliberately deciding where to go and more dependent on systems predicting what would keep them engaged.

The consequences have produced a more complicated view of the original promise. Research does not support a simple claim that social media causes a loneliness or mental-health crisis, since effects vary considerably by person, behaviour, platform, and context. A 2026 meta-analysis covering 81,344 adolescents and young adults found only a small relationship between time spent on social media and loneliness, but a substantially stronger association between problematic social media use and loneliness. [5] (Social Psychiatry and Psychiatric Epidemiology)

The U.S. Surgeon General has similarly concluded that social media can provide real benefits while also presenting meaningful risks to young people, and that the available evidence is not sufficient to conclude that social media is safe for children and adolescents. [6] (U.S. Department of Health and Human Services) Research on political polarization has identified another set of concerns, including the ways platform design, algorithms, partisan selection, and the kinds of messages that spread online can contribute to political division. [7] (PubMed)

None of this has caused society to abandon social media. Instead, the conversation around it has become more complicated. The question is no longer simply whether connecting everyone is good. It now includes questions about how platforms are designed, what recommendation systems amplify, how children should use them, how much time people want to spend inside feeds, and what responsibilities technology companies have for the environments they create.

People can continue valuing what a technology gives them while becoming more conscious of what it changes. AI may eventually produce its own version of that negotiation.

AI is changing the journey to an answer

Search engines transformed the internet by making an enormous collection of websites navigable. For most of the web’s history, even a relatively simple question required some movement through that system. A person searched, chose a result, opened a page, evaluated what they found, and often continued somewhere else.

Generative AI changes that relationship because an AI assistant can increasingly perform much of that navigation on the user’s behalf. It can interpret information, combine material across sources, and return a synthesized response. The underlying websites may still be available, but visiting them is no longer always necessary to complete the task.

The major AI and search companies describe this as a benefit. OpenAI introduced ChatGPT search as a way to provide fast answers with links to relevant sources, combining conversational interaction with current information from the web. [8] (OpenAI) Google describes AI Mode as a way for people to ask longer and more complicated questions, receive synthesized responses, continue through follow-up questions, and explore supporting links when they want to go deeper. Google says AI Overviews have increased search activity for the kinds of queries where they appear. [9] (Google)

Early behavioural evidence suggests that this changes how often people continue into the wider web. Pew Research Center analyzed the browsing behaviour of 900 U.S. adults in March 2025 and found that when a Google search included an AI summary, users clicked a traditional search result on 8 percent of visits. When no AI summary appeared, they clicked a result on 15 percent. Only 1 percent of visits to pages containing an AI summary resulted in a click on one of the sources cited inside the summary, and users were also more likely to end their browsing session after receiving an AI-generated answer. [10] (Pew Research Center)

The study reflects an early stage of AI search and should not be treated as a permanent description of how people will behave. Google has continued redesigning AI search to make sources more visible and says its own testing shows that some of those changes increase visits to supporting websites. [11] (Google)

Even with those qualifications, the underlying change is significant. The internet is becoming capable of satisfying more informational needs before a person enters the destination from which the information came.

Sometimes the answer is all we need

There is no reason to romanticize the work AI is removing. A large amount of internet use is functional. Someone may want to know whether a store is open, which tax form to complete, whether two products contain the same ingredient, how to reset a device, what a technical term means, or which train gets them somewhere fastest. In situations like these, moving through several websites is not necessarily valuable and can often be irritating.

Years of search optimization also created an enormous amount of content whose primary purpose was to capture queries. A simple question could lead to an unnecessarily long article surrounded by advertisements, pop-ups, affiliate links, and several paragraphs written largely for a search engine. AI is well suited to collapsing that experience into something more useful.

There are also situations in which synthesis is itself the service people want. Someone comparing insurance policies may prefer a system that can read several documents and explain the differences. Someone approaching an unfamiliar subject may benefit from a clear overview before deciding what deserves more attention.

The case for an exploratory internet therefore cannot rest on the idea that people should continue doing work that a machine can perform more efficiently. The more interesting question is whether finding the answer was always the only thing people were doing while they moved through the web.

Sometimes looking is part of the experience

Information science has long distinguished between retrieving a known piece of information and more exploratory forms of information seeking. Exploration becomes useful when someone does not yet understand a subject well enough to formulate the right question, when the goal is broad, or when useful information may emerge unexpectedly during the process.

Serendipity forms part of this behaviour. A systematic review of 207 studies on serendipity in human information behaviour describes unexpected information discovery as an important way people encounter and acquire knowledge. Earlier research on web browsing similarly argued that information systems should sometimes support people’s ability to explore rather than attempt to replace the process entirely. [12] (Journal of Documentation; International Journal of Human-Computer Studies)

More recent research gives the idea a behavioural dimension. A 2026 study involving 799 adults allowed participants to freely explore a hypertext on a historical subject and tracked their clicking and reading behaviour. The researchers found that the depth of engagement with additional information, measured through reading time, was associated with curiosity. [13] (Journal of Experimental Psychology: Applied)

The point is not that clicking links is inherently better than asking an AI. It is that information seeking can itself form part of the experience. People do not always enter an information environment with a perfectly formed question and leave once a fact has been retrieved.

Someone reading about an election may become interested in the history of a country, while someone researching a camera may discover a photographer whose work changes what they want from the camera in the first place. A shopper browsing a clothing store may encounter a designer they did not know, and a reader may follow an unfamiliar reference in an essay into an entirely different subject. In each case, the eventual discovery was not fully contained inside the original intention. The person’s understanding developed while they were looking.

AI can support curiosity too

There is a danger in turning this argument into another simple opposition between technology and human curiosity. There is no evidence that using AI inevitably makes people passive or incurious. In some cases, AI may make exploration easier by helping people formulate better questions, follow unfamiliar ideas, or approach subjects that previously felt inaccessible.

The Reuters Institute’s 2026 Digital News Report provides a useful complication. Ten percent of respondents across the markets it studied said they used AI chatbots for news each week, up from 7 percent the previous year, but those users were not unusually disengaged from information. Reuters described them as highly engaged and digitally sophisticated. They were more likely than the general population to use multiple news outlets, search engines, messaging apps, video platforms, individual journalists, and alternative sources. Thirty-eight percent fell into Reuters’ most interested “news lover” segment, compared with 22 percent of respondents overall. [14] (Reuters Institute)

Their reasons for using AI were also revealing. Forty-two percent said they used chatbots to ask follow-up questions, suggesting that AI can support inquiry as well as replace parts of it.

The more consequential question is therefore not whether AI and curiosity can coexist. It is what happens when AI becomes the default interface through which people encounter information. A highly curious person may use an AI response as the beginning of a journey, while another person may find that the response is enough. The same system can support both behaviours, and the balance between them may change as AI becomes more deeply embedded in everyday internet use.

Getting an answer and exploring may become separate activities

For most of the internet’s history, finding an answer and navigating the web were closely connected because information retrieval usually required some degree of exploration. AI begins to separate those activities.

One form of internet use is primarily functional. A person wants to know something, compare something, understand something, or complete a task, and AI can increasingly handle much of the work required to reach that outcome. Another form begins without such a narrow endpoint. A person enters a publication, store, community, archive, creator’s work, or specialist website because they want to see what is there.

Those behaviours already coexist in other parts of culture. Streaming made recorded music available almost instantly without eliminating the desire to experience music live. Online shopping made purchasing more efficient without eliminating stores people visit partly for the experience of browsing. Industrial food made meals easier to produce without eliminating cooking or the cultural value attached to ingredients, restaurants, and food provenance.

These examples do not prove that the internet will follow the same path, but they suggest that a more convenient way of completing an activity does not always eliminate the less efficient version. Instead, the purpose of the older behaviour can change.

If AI becomes the easiest way to answer many questions, manually moving through websites may become less necessary for routine information. At the same time, experiences where people want to browse, discover, compare perspectives, or immerse themselves in a subject could become more distinct.

The result could be an internet with two increasingly recognizable modes: an answer layer that helps people reach an outcome quickly, and an exploratory layer made up of places people enter because looking around contributes something to the experience.

This could change what makes a website valuable

The immediate business concern around AI and websites is traffic, and that concern is justified. If an AI system can satisfy an informational need without sending someone to the underlying source, websites that depend heavily on search referrals face a structural problem. Pew’s browsing research already shows substantially lower click-through when AI summaries appear in Google results. [10] (Pew Research Center)

AI may also expose a deeper weakness in many websites. If a website exists mainly to hold information that can be accurately extracted, summarized, and delivered somewhere else, the website itself may no longer be the most useful place to consume that information.

The reason to visit therefore has to extend beyond access to information. For a publisher, that could come from distinctive reporting, writers, archives, data, tools, communities, or a recognizable editorial worldview. For a retailer, it could come from helping people understand a category, compare possibilities, browse products in context, learn from experts, or discover things they did not know to search for. For a B2B company, it could come from original research, demonstrations, customer evidence, expert thinking, and deeper explanations of complicated problems. For a brand, it could come from a distinctive world of ideas, products, people, stories, and experiences that rewards someone for spending time inside it.

These experiences do not need to compete with AI at producing a faster answer. Their value comes from giving people a reason to continue exploring when obtaining an answer is no longer the only reason to visit.

What this means for digital leaders

The rise of AI does not remove the need for websites to be understandable to machines. Information should be structured clearly enough for search engines, AI systems, commerce agents, and other intermediaries to interpret it. Brands will increasingly operate in environments where customers encounter their information without entering their website first.

That addresses only one side of the problem. Digital leaders should also think more carefully about what happens when someone chooses to enter the owned experience and whether the website rewards that decision.

Separate information from experience

Not every piece of information needs to become an elaborate destination. Product specifications, policies, definitions, support content, and straightforward answers should be easy for both people and machines to understand. Richer experiences make more sense around the subjects, decisions, tools, stories, communities, and interactions where spending additional time creates additional value for the visitor.

Design for people who do not know exactly what they want

Many websites are designed around explicit intent, such as finding a particular product, navigating to a known category, reading a specific article, or completing a task. Exploration requires the website to accommodate people whose intention is less defined.

That can mean creating useful paths between adjacent ideas, products, people, stories, and subjects so that someone can discover more without being pushed into an endless stream of recommendations. The objective is not to maximize the amount of content consumed. It is to make the time someone chooses to spend exploring worthwhile.

Create value that cannot be completely pre-consumed

AI is increasingly capable of extracting and summarizing information that already exists on a website. That makes other forms of value more important, including original tools, interactive experiences, communities, evolving datasets, distinctive visual environments, expert relationships, and bodies of work that are more useful when experienced in context.

The opportunity is not to make information difficult for AI to access. Businesses will still benefit from making their information understandable and available across external systems. The opportunity is to create additional reasons for people to experience the source itself.

Give people reasons to return directly

A website becomes more defensible when people remember that it exists and expect to find something worthwhile when they return. That expectation can come from recurring research, changing product assortment, strong editorial judgment, community, useful tools, entertainment, expertise, or a body of work that develops over time.

Direct relationships are unlikely to replace discovery through search, social platforms, creators, marketplaces, or AI. They can, however, give businesses a relationship with audiences that does not depend entirely on those intermediaries.

The internet after the answer

Major technological shifts often produce predictions that an existing behaviour is about to disappear. Some behaviours do disappear, but others survive because the new technology changes what the older activity is for.

Industrial food did not eliminate cooking, just as streaming did not eliminate live music and online retail did not eliminate physical shopping. The behaviours that remained were not always the most efficient ways to accomplish the functional task. Cooking takes longer than heating a prepared meal, attending a concert requires considerably more effort than streaming a song, and wandering through a store is less efficient than ordering a known product online. They continue because efficiency is only one part of what people value in those experiences.

AI now offers the internet an extraordinary form of efficiency. It can read, compare, summarize, explain, translate, recommend, and increasingly act. For many everyday questions, asking an AI may become easier than navigating the web ourselves, and there is little reason to resist that simply for the sake of preserving an older way of doing things.

It may nevertheless be too early to conclude that destinations become irrelevant simply because answers no longer require them. Human beings do not only seek information to resolve questions. We also explore because we are curious, enjoy discovering things, want to understand where ideas came from, encounter different perspectives, or simply do not yet know what we are looking for.

The first phase of the AI internet is understandably focused on how much of this work machines can do for us. A later phase may involve learning which parts of the experience we still want to do ourselves.

If that rebalancing happens, the web will not need to compete with AI by producing faster answers. Its opportunity will be to give people somewhere worth going when the answer is not enough.

The Idea

NewBrains proposes two related ideas from this analysis.

The answer layer describes the growing role of AI systems in retrieving, interpreting, combining, and delivering information without requiring people to navigate each underlying source themselves.

The exploratory internet describes the websites, communities, publications, stores, creators, tools, and other digital destinations people choose to enter when exploration contributes to the value of the experience.

These modes are not opposites and are likely to coexist. AI can make routine information retrieval dramatically more efficient while also making the distinction between getting an answer and exploring a subject more visible.

If that distinction grows, the strategic question for websites changes. The challenge will not only be whether information can be found, ranked, cited, or interpreted by AI. It will also be whether there is something worth coming to see.

Sources

1. Poti, Jennifer M., Michelle A. Mendez, Shu Wen Ng and Barry M. Popkin
Is the Degree of Food Processing and Convenience Linked With the Nutritional Quality of Foods Purchased by US Households? American Journal of Clinical Nutrition, 2015. Analysis of more than 1.2 million products and U.S. household food purchasing patterns. (PubMed)

2. U.S. Department of Agriculture, Economic Research Service
Organic Agriculture. Updated January 2026. Data on the growth of U.S. organic agriculture, certified operations, distribution, and consumer sales. (USDA)

3. International Food Information Council
2026 Food & Health Survey. Findings on food purchasing priorities, healthfulness, convenience, and consumer attitudes toward food processing. (IFIC)

4. Pew Research Center
The Strength of Internet Ties. 2006 research examining how internet use affected social networks, relationships, information seeking, and community. (Pew Research Center)

5. Feng, Hsin-Pei et al.
Exploring the Relationship Between Social Media Use and Loneliness in Adolescents and Young Adults: A Systematic Review and Meta-analysis. Social Psychiatry and Psychiatric Epidemiology, 2026. Meta-analysis of 59 studies and 81,344 participants. (Springer Nature)

6. U.S. Surgeon General
Social Media and Youth Mental Health. Advisory examining evidence on the benefits and potential harms of social media use among children and adolescents. (U.S. Department of Health and Human Services)

7. Van Bavel, Jay J. et al.
How Social Media Shapes Polarization. Trends in Cognitive Sciences, 2021. Review of social, cognitive, technological, and algorithmic mechanisms associated with political polarization. (PubMed)

8. OpenAI
Introducing ChatGPT Search. October 2024. Description of ChatGPT’s integration of conversational AI with current information and web sources. (OpenAI)

9. Google
AI in Search: Going Beyond Information to Intelligence. May 2025. Description of AI Mode, AI Overviews, complex querying, follow-up questions, and changes in search behaviour. (Google)

10. Pew Research Center
Google Users Are Less Likely to Click on Links When an AI Summary Appears in the Results. July 2025. Analysis of browsing data from 900 U.S. adults examining click-through behaviour following Google searches with and without AI summaries. (Pew Research Center)

11. Google
How AI Mode and AI Overviews Help You Explore the Web. 2026. Description of changes intended to increase the visibility of original websites and sources within generative search experiences. (Google)

12. Makri, Stephann et al.; Beale, Russell
Research on serendipity and exploratory information seeking, including a systematic review of serendipity in human information behaviour and earlier research on supporting exploration during web browsing. (Journal of Documentation; International Journal of Human-Computer Studies)

13. Schumacher, Aki et al.
From Clicks to Curiosity: Exploring Self-Directed Information Seeking as a Behavioral Manifestation of Curiosity. Journal of Experimental Psychology: Applied, 2026. Study of 799 adults examining self-directed exploration, reading behaviour, and curiosity. (American Psychological Association/PubMed)

14. Reuters Institute for the Study of Journalism
Digital News Report 2026 and Emerging Uses of AI Chatbots for News and What It Means for Journalism. Research across international markets on AI chatbot adoption, information-seeking behaviour, news engagement, and click-through to original sources. (Reuters Institute)