A New Era of AI Search
The internet is entering a new era in which search is no longer defined simply by typing keywords into a box and choosing from a list of links. Artificial intelligence is changing the basic relationship between people and information. Instead of asking users to search, open several websites, compare information and construct their own answer, AI-powered search systems increasingly attempt to understand the question, investigate multiple sources, synthesize the information and present a direct response. The shift is turning search from a system for finding webpages into an increasingly conversational system for understanding the web.
Google’s transformation illustrates how rapidly this change is taking place. At its 2026 I/O conference, Google described its latest developments as a “new era for AI Search,” introducing a redesigned search experience built around AI. The company said its AI Mode had surpassed one billion monthly users and that queries had more than doubled every quarter since launch. Google also introduced a new AI-powered Search box capable of working with text, images, files, videos and Chrome tabs, allowing users to provide substantially more context than a traditional keyword query.
This represents a fundamental change in how a search query can work. Traditional search generally begins with a small number of keywords and returns a collection of documents that the user must evaluate. AI search can instead interpret a longer, conversational request and break a complicated question into multiple subtopics. Google’s description of AI Mode says it can search several aspects of a question simultaneously, while retaining conversational context for follow-up questions. The result is closer to an interactive research session than a conventional results page.
OpenAI’s development of ChatGPT search has contributed to the same transition from keyword search toward conversational discovery. OpenAI introduced ChatGPT search as a way to combine natural-language conversation with current web information, allowing users to ask questions, receive answers supported by web sources and continue with follow-up questions without having to restart the search process. The company has emphasized links to original sources so users can examine the underlying material themselves.
Perplexity has approached the same transformation from the opposite direction, building an AI-native search experience around research and synthesis. Its Pro Search system describes a process in which an AI model conducts multiple searches, gathers information from different types of sources and synthesizes those materials into a coherent response with citations. This reflects an emerging expectation that a search engine should not merely retrieve information but perform part of the research process for the user.
The most important change may therefore be the evolution of the search query itself. People are becoming less constrained by the traditional requirement to guess which keywords will produce the right webpage. AI search allows questions to be expressed in ordinary language, including detailed instructions, preferences, context and follow-up questions. Google has explicitly redesigned its Search interface around this behavior, while ChatGPT and Perplexity have made conversational interaction central to their search experiences.
The consequences for websites and publishers are potentially profound. For decades, search engines largely functioned as gateways: a user entered a query, received links and then visited an external website. AI search can instead provide a synthesized answer before the user ever clicks a link. This creates a new tension between giving users faster answers and preserving the flow of traffic that supports much of the open web.
Recent industry data illustrates the scale of that change. A July 2026 analysis reported by TechCrunch, citing Similarweb data, found that Google’s AI Overviews appeared in 43% of searches analyzed, up from 15% a year earlier. The same analysis reported growth in AI Mode visits and noted concerns about publishers receiving fewer referrals when information is consumed directly inside AI-generated responses. These figures describe a rapidly changing search environment, although individual search behavior and results can vary considerably by query, country and device.
Google is now experimenting with economic mechanisms around this new information ecosystem as well. In September 2026, reports emerged about a Google “AI contribution pilot” that would compensate participating publishers when their content contributes to AI-generated answers in products including Gemini, AI Overviews and AI Mode. The reported experiment is significant because it points toward a possible future in which the value of a publisher’s content is measured not only by clicks but also by how much that content contributes to an AI answer. Google has confirmed that the initiative is an early-stage pilot.
This could eventually change the meaning of search-engine optimization itself. In the traditional search economy, publishers compete for visibility among ranked webpages. In AI search, visibility can take another form: being selected as a source from which an AI system constructs its response. That means publishers may increasingly need to consider whether their information is easy for AI systems to discover, interpret, attribute and cite. The industry has consequently begun discussing concepts such as “answer engine optimization” alongside conventional SEO, although the precise practices and standards are still evolving.
The transformation is also moving search from passive retrieval toward active assistance. Google’s 2026 announcements included information agents designed to operate in the background, monitoring the web for changes related to a user’s interests and delivering synthesized updates. The company gave examples involving apartment searches and product developments, illustrating a model in which people do not necessarily need to perform the same search repeatedly. Instead, an AI system can continue watching the information environment on their behalf.
That development could be one of the biggest conceptual changes in the history of search. The classic search engine waits for a query. An agentic search system can continue working after the original question has been asked. The distinction is important because information itself is increasingly dynamic. News changes, prices move, products appear and disappear, regulations are updated and events develop over time. A system capable of monitoring these changes can transform search from a one-time transaction into an ongoing information service.
AI search is also becoming multimodal. Google’s new Search experience allows users to work with combinations of text, images, files, videos and browser content. This matters because many real-world questions cannot be expressed efficiently through keywords alone. A photograph, document, screenshot or video can contain information that would be difficult to describe manually. AI systems can potentially use these different forms of information together, making search more closely resemble human investigation.
Another emerging characteristic is personalization. Google has expanded Personal Intelligence within AI Mode, allowing users in many countries and territories to connect services such as Gmail and Google Photos when they choose to do so. The stated objective is to make search understand not only the information available on the internet but also relevant aspects of a user’s own context. Such systems could eventually answer questions that require both public information and personal information, although privacy, security and user control become much more important as that boundary expands.
The increasing power of AI search also makes verification more important, not less. A conventional search result generally tells users which document they are looking at, while an AI-generated answer presents a synthesized interpretation. Even when citations are supplied, users may not immediately know which statement came from which source or whether the AI has interpreted the evidence correctly. Google itself describes AI-generated search content as experimental in some of its documentation, and its AI Mode documentation says that the system can fall back to web links in situations where confidence in an AI response is insufficient.
This creates an important paradox. AI search is designed to reduce the effort required to obtain information, but the easier an answer becomes to consume, the easier it can also become to accept without examination. The most valuable future search experience may therefore not be one that simply produces the shortest answer. It may be one that combines synthesis with transparent sourcing, distinguishes established facts from uncertainty and makes it easy for users to investigate the original evidence.
For journalism, the implications are particularly significant. News organizations have historically relied on search engines as important distribution channels, with readers moving from search results to individual stories. If AI systems increasingly summarize news directly, publishers could gain exposure through citations while simultaneously losing direct visits. The emerging experiments around AI-content attribution and compensation suggest that the technology industry and publishing industry are still searching for a sustainable economic relationship.
For businesses, AI search changes another fundamental question: how customers discover products and services. Instead of searching separately for specifications, reviews, prices and alternatives, a consumer may increasingly ask an AI system to compare those factors in a single conversation. Google’s search strategy already includes agentic shopping capabilities, while its advertising developments are being adapted to AI-driven search environments. The commercial battle is therefore moving beyond ranking webpages toward influencing how AI systems understand products, brands and user intent.
The competitive landscape is consequently expanding. Google brings decades of search infrastructure, a vast index and the Gemini model family. OpenAI brings a conversational interface that already functions as a destination for millions of information-seeking interactions. Perplexity has focused heavily on AI-native research and source synthesis. Other technology companies are also developing AI assistants and agentic systems. The competition is no longer simply about which company has the best list of webpages; it is increasingly about which system can understand intent, retrieve reliable information, reason over it and help the user accomplish something.
At the same time, the open web remains essential to AI search. AI systems need fresh information, original reporting, specialist knowledge, reviews, public records and countless other forms of human-created material. If users increasingly consume synthesized answers without visiting the sources that produced the underlying information, the economics of creating that information could come under pressure. The future of AI search will therefore depend partly on whether a sustainable ecosystem can emerge between AI companies, publishers, creators and websites.
The phrase “a new era of AI search” is consequently more than a description of a new interface. It represents a transition in the role of the search engine itself. The search engine of the past primarily helped people locate information. The emerging AI search engine attempts to understand the question, investigate the available information, synthesize an answer, maintain context and increasingly take action.
The transition is still underway, and many of its long-term consequences remain uncertain. Search interfaces, business models, publisher relationships, advertising systems and standards for attribution are all being redesigned at the same time. What is already clear is that the traditional boundary between “search engine” and “AI assistant” is becoming increasingly difficult to maintain.
The next generation of search may therefore feel less like searching a database and more like working with an intelligent research partner. Users will still need to ask questions, examine sources and exercise judgment, but the machinery between the question and the answer is becoming dramatically more sophisticated. As Google, OpenAI, Perplexity and others compete to define this new model, the central contest is moving from who can retrieve the most links to who can help people understand and act on the world’s information most effectively.
