Pew: 60% of Americans Now Read AI Summaries in Search
On 17 June 2026, the Pew Research Center published Americans and AI 2026: Chatbots, Smart Devices and Views on Impact, a survey of 5,119 US adults. The headline finding for search is that around 60% of US adults have now read AI-generated summaries in search results, a level that puts AI answers firmly into mainstream behaviour rather than early adoption.
Pew is a stronger source for this kind of demand-side data than the tool-company surveys usually quoted on AI search, which makes the figures worth treating as a reference point rather than marketing.
What the survey found
The numbers describe a population that has largely tried AI tools, led by search summaries and chatbots:
- Around 60% of US adults have read AI-generated summaries in search results. Men were slightly more likely than women to report reading them (63% versus 57%), and adults aged 65 and over were the least likely age group.
- 49% have used an AI chatbot, up substantially from Pew’s 2024 measurement, with roughly one in four using one daily. ChatGPT leads usage at 44%, ahead of Google Gemini and Microsoft Copilot.
- 40% use chatbots for information lookup, the single most common use of the technology.
Adoption is not matched by confidence. Around 63% of respondents say AI is moving too fast, and 71% worry it will make their personal data less secure. Only a minority believe AI will benefit society.
Why the date matters
The survey was fielded between 17 and 23 February 2026. The report is therefore newly published rather than newly collected: it captures sentiment from early 2026, not June. Given how quickly AI search adoption has moved, the true current figures are likely higher still, which makes the 60% and 49% numbers a conservative floor rather than a ceiling.
What this means
Two practical implications follow for publishers.
First, AI-surface visibility is no longer a fringe concern. When a majority of US adults read AI summaries in search, content that is not retrievable or citable in those summaries is invisible to a large and growing share of the audience, even where traditional rankings hold. This is the demand-side case for treating generative engine optimisation and AI visibility measurement as part of core SEO rather than an experiment.
Second, the usage-versus-trust gap is itself a signal. People are using AI answers far faster than they trust them. That favours sources users recognise and can verify: named authors, transparent sourcing, and credible brands. The same E-E-A-T signals that earn citations also address the trust deficit the data exposes. Being the source an AI answer cites, and that a sceptical reader is willing to believe, is the durable position.
Sources
More news
-
People Also Ask Answers Hit 97% AI-Generated, After April's Slump to 11%
AlsoAsked measured 97% of English PAA answers as AI-generated in early September, up from 11% in April. The same curve has reversed twice this year.
-
ChatGPT Now Cites Twice as Many Sources, and Listicles Lost Half Their Share
GPT-5.6 searches more and scopes searches to specific domains. Listicles and comparison pages lost citation share, but the pool they compete in doubled.
-
Shopify Finishes a 10-Month Expansion of Its Theme Performance Docs
Shopify has concluded a rewrite of its theme performance best practices, including a page on speed apps that cheat testing tools, and built the docs for agents.