Gen AI inside existing search engines overtakes standalone gen AI

Passive vs. standalone gen AI

Common examples of where gen AI technology will be used passively include search, e-commerce, social media, and online news. This usage of gen AI inside existing apps contrasts with what we may term the “traditional” usage of standalone gen AI apps, such as ChatGPT or Gemini, which users open on their devices and use specifically to create an output—be it text, image, code, or another type.

With passive gen AI use, the technology is an embedded, essential but not overt capability within another application. The user is not explicitly using gen AI, but this technology is core to the experience. For example, gen AI may be used to synthesize numerous responses from a search; to summarize thousands of individual product reviews; or to create content disseminated via social media or online news. 

Deloitte estimates that comparing the usage of all passive gen AI apps relative to all standalone, proactively used apps would show the former already being notably more popular in 2025 as well. In the UK market, where emerging products are often early to launch, Deloitte UK’s research found that as of mid-2025, about three-quarters of respondents had ever used one of four types of passive gen AI applications²—notably ahead of the 47% of respondents who had ever used any dedicated, standalone gen AI app.

Another metric for emerging applications is comparing usage at any time in its relatively short history. Passive gen AI applications were first launched in the United States in May 2024 with the introduction of search summaries,³ and rollout into additional markets was announced in November of that year⁴—almost two years after the launch of the first popular standalone gen AI apps in late 2022. Despite standalone apps having this lead, we forecast that by mid-2026, more adults will have generated a search overview (72%) than those who have used a standalone gen AI tool at any time (61%) (figure 2).

What can we learn from user preferences for passive search?

Search, social media, and e-commerce are already among the most frequently used digital applications. There are over 15 billion searches undertaken every day. On average, users spend over two hours on social media daily.⁵ E-commerce sales in Q1 2025 alone in the United States totaled $300 billion.⁶ Users may be more likely to use gen AI capabilities within a familiar search tool rather than search within an unfamiliar, novel gen AI chatbot.

In 2026, questions about gen AI’s impact on the viability of the search business model may continue, but there may also be questions about the impact of gen AI-enhanced search on the popularity of standalone gen AI tools such as ChatGPT or Synthesia.⁷ According to Deloitte’s research, the most common workplace application for gen AI is search. It may be that some users who currently search within standalone gen AI apps move back to mainstream search applications.

The pace at which passive gen AI has overtaken standalone gen AI is impressive and, perhaps, predictable. Standalone gen AI is both presented and perceived as novel and relatively experimental. It requires skill and persistence: a disappointing outcome may result from a poor prompt rather than a flawed model, and the remedy is to re-prompt.⁸ It may be the user’s prompt-engineering skills that are blamed rather than the product. Passive gen AI should be lower friction if it’s an incremental capability that is seamlessly integrated into an existing mainstream digital application, be it search engine, e-commerce site, social media app, or office productivity tool. There is rarely a need to try again. The technology is less overt, the experience more familiar, and, as such, the demand is greater because it’s more accessible. The application of gen AI to create a summary of search results automates and completes a task that many users otherwise would have done manually—that is, click on and read multiple links to formulate a personal summary, a chore that is eliminated by the AI summary. The integration of gen AI into an existing application is akin to one-touch checkout, including payment, integrated into e-commerce sites, or facial-recognition authentication incorporated into a consumer banking app.

Adoption trends across generations

Passive AI’s accessibility is evident in the rapid adoption of search summaries among older age groups, who may be less inclined to master new standalone tools. As of mid-2025, boomers were hesitant about standalone gen AI. Deloitte’s research found that only 20% of boomers had ever used any generative AI tool—despite an awareness rate of 58%. By contrast, almost four times as many (76%) members of Generation Z had used a gen AI tool in 2025 (figure 3). However, adoption of search overviews was 75% higher among boomers—at 35%—relative to any standalone tool. 

Bottom line: Passive AI usage has market implications for gen AI

Gen AI is one of the most important technologies of its time, but its fullest potential may only be realized when it’s deployed additionally as a discreet, yet integral, capability within existing, mainstream applications.  

Many of today’s most important technologies began as standalone capabilities, often within dedicated devices. It was not long ago that GPS, or sat-nav referred to a physical appliance so useful that users took it on work trips and vacations. This functionality was then integrated into smartphones and their applications. Now, satellite navigation is integrated into myriad applications beyond route finding—its usage is vital, ubiquitous, and largely in the background.

Gen AI often improves existing applications, even if it may not make them perfect. It can summarize search results, and while it may introduce errors when doing so, in many cases this may not matter. Further, users may trade the simplification of the search process enabled by gen AI for the errors that may be introduced by the technology’s inherently probabilistic approach.

For the many standalone gen AI app owners in the market, a core question to address in 2026 will be to consider choosing between focusing on embedding their tools’ capabilities within another application or to remain as a standalone interface—the latter approach generating higher revenue per user but potentially lower adoption. A few players will be able to do both, but for the remainder, a choice may need to be made.