Generative AI Statistics 2025: The Latest Trends & Insights

2025 marks a clear turning point for generative AI. The year the technology moved decisively from early-adopter novelty to everyday utility. It now shapes how we search, learn, create, and make decisions online. 

The sudden increase in traffic is just the tip of the iceberg. What’s even more interesting is who is using AI, and how. Once dominated by younger digital natives, the audience is rapidly broadening. Users aged 18–34 still form the largest share, but their share has dropped from 61% to 53%, as older demographics increasingly incorporate AI tools into their daily workflows. 

As detailed in Similarweb’s 2025 Generative AI Landscape report, the shift is unmistakable: generative AI has moved from a curiosity used by early adopters to a daily tool shaping decisions, discovery, and behaviour across the web.

How Gen AI platforms overtook the web

1. Adoption at scale: Where Gen AI stands now

Generative AI adoption has accelerated at a pace few technologies have matched. Over the past year, monthly web visits to Gen AI platforms climbed to 7 billion, while mobile apps surged with 1.9 billion downloads, signalling a shift from experimentation to habitual, everyday use. 

At the center of this momentum is ChatGPT, which now commands nearly 80% of global Gen AI traffic and has risen into the world’s top five websites, which underscores its dominance even as platforms like Gemini, Perplexity, and Claude carve out fast-growing niches. Together, these trends reveal a maturing landscape defined by rapid adoption, clear category leaders, and an increasingly competitive ecosystem.

2. A new platform hierarchy emerges

The generative AI landscape is entering a new era defined by both clear leadership and growing fragmentation. At the top of the hierarchy sits ChatGPT, which accounts for nearly 80% of global Gen AI visits.

But beneath this dominance, the market is diversifying quickly. Google’s Gemini is gaining momentum through tight integration with Search and Workspace, while emerging players like DeepSeek, Claude, Perplexity, and Grok are carving out niche audiences seeking advanced reasoning, open-source transparency, or privacy-focused alternatives. The introduction of Google’s AI Mode, which blends search and generative responses, is accelerating this fragmentation by reshaping how users navigate the web.

3. How users move between web and mobile AI

As generative AI becomes part of everyday digital behaviour, users are increasingly moving fluidly between web and mobile, but mobile is where the deepest engagement is taking shape. While web platforms now reach that signals a shift from casual exploration to embedded, habitual use.

Within mobile, AI apps are beginning to split into two clear tiers: 

• Scale platforms
• Specialised tools

ChatGPT dominates with 41 million monthly active users and a 33% stickiness rate, far outpacing other competitors, which is a sign that it has become a daily-use utility for millions. Meanwhile, apps like Perplexity, Copilot, Gemini, Claude, and Grok attract smaller audiences and much lower stickiness rates (5%–17%), reflecting more occasional, task-specific use.

Together, these shifts illustrate a maturing ecosystem. The web drives reach, mobile drives routine, and users are increasingly gravitating toward platforms that offer both breadth and depth, or niche tools that deliver targeted value.

What really happens inside Gen AI sessions

1. Shifting intent behind AI prompts

Gen AI sessions reveal a fundamentally different pattern of user intent compared to traditional search. While Google searches remain heavily informational (64%), ChatGPT interactions are far more varied, with only one in five prompts focused on information retrieval. Instead, users turn to AI for creative ideation (32%), practical guidance (24%), and a growing set of technical or problem-solving tasks (22%).

This shift also shows up in how people engage. ChatGPT prompts average 60 words, compared to Google’s 3.4-word queries, reflecting a move toward richer, more contextual inputs designed to shape personalised responses. As a result, Gen AI sessions are becoming more conversational, iterative, and exploratory, blending search, creation, and decision-making into a single flow. 

2. The new AI citation ecosystem

Citations are becoming a defining feature of how Gen AI platforms source and surface information. Our citation analysis shows that ChatGPT’s use of citations has grown steadily throughout 2025, rising from 0.9% in early spring to 5.9% by October, reflecting a shift toward answers grounded in live web data rather than solely pre-trained knowledge.

This trend signals an expanding ecosystem of discovery in which publishers, news outlets, and user-generated content platforms, especially Reddit, play an outsized role.

Referral behaviour highlights a second major shift. While overall referral volume from Gen AI platforms grew sharply earlier in the year, it has since begun to plateau, and referral rates remain modest. Google Search drives referral rates of 17–19%, whereas Google’s AI Mode averages just ~2%, and ChatGPT sends relatively few outbound clicks.

Yet when users do click through from AI platforms, their visits are significantly higher quality: longer time on site, more pageviews, and stronger conversion rates than traditional search traffic.

3. The new discovery landscape

Discovery is rapidly evolving as generative AI becomes a central gateway to information. One major shift is the rise of AI-powered interfaces embedded directly into users’ existing digital habits. Google’s AI Mode, the fastest Gen AI platform to reach 100M US visits, blends conversational answers with traditional search, indicating a future where discovery happens inside hybrid environments rather than across standalone tools.

At the same time, AI browsers like ChatGPT, Atlas, and Perplexity Comet are redefining how people interact with the web, allowing users to query, analyse, and act on what appears on their screen in real time. Early adoption spikes suggest that browsing itself is becoming an AI-driven experience.

AI is also reshaping how brands appear in discovery pathways. LLMs increasingly favour news publishers, authoritative content, and user-generated platforms like Reddit, often surfacing deep, niche pages rather than top-level URLs. As a result, visibility is shifting from traditional ranking positions to being part of the “knowledge scaffolding” that AI models use to build answers.

The new AI search playbook

1. What brand visibility looks like in AI search

Unlike search engines that send users directly to external sites, Gen AI platforms synthesise information and surface brands only when they appear as trusted, repeated signals across the web. This new dynamic makes visibility, not traffic, the defining KPI for AI-era discoverability.

Brands with well-structured, authoritative, and frequently referenced content rise to the top of AI conversations. In retail, for example, leaders like Target and Walmart appear in more than half of relevant AI responses, while in travel, established platforms such as Airbnb, Booking.com, and Expedia dominate brand mentions. These patterns reveal how AI models prioritise brands that consistently appear across reputable sources, UGC platforms, and deep content pages.

At the same time, the nature of AI-generated citations is evolving. ChatGPT continues to increase the share of responses that reference external sources, drawing heavily from publishers, review sites, and platforms like Reddit. This creates a discovery ecosystem where brand influence depends not just on owned content, but on how frequently a brand is referenced in the broader web, from expert reviews to community discussions.

2. Structuring content for AI discovery

As generative AI reshapes how users discover information, the structure and depth of brand content have become critical to being surfaced in AI-generated answers. Unlike traditional search engines that favour top-level, keyword-optimised pages, LLMs pull heavily from deep, highly specific URLs, with a third of ChatGPT citations coming from pages three folders deep and significant volume from even deeper paths. This means brands must prioritise detailed, intent-rich content that answers precise user questions, especially long-tail, conversational prompts that LLMs excel at interpreting.

Clarity and structure matter just as much as depth. AI models more easily interpret pages with:

• Clearly labelled sections
• Bullet-point summaries
• Distinct product or feature attributes

The report highlights how Walmart’s product detail pages, with a dedicated “About this item” section, outperform less-structured competitors like Temu in AI visibility. Well-organised content creates a reliable blueprint for LLMs to extract and synthesise information.

But being the source isn’t the only path. LLMs also favour authoritative third-party domains, including publishers, expert reviews, and user-generated platforms such as Reddit. To optimise fully for AI visibility, brands must ensure not only that their own content is structured for machine understanding, but that they show up across the ecosystem of sources AI trusts.

In the AI era, content optimisation expands beyond SEO. It’s about creating information that is machine-readable, deeply specific, and repeatedly validated across the web, the signals LLMs use to decide which brands deserve to be part of the answer.

3. Signals shaping the future of Gen AI

The next wave of AI-driven discovery is already taking shape, and several clear signals point to where the landscape is heading. First, AI is becoming embedded in everyday browsing, with Google’s AI Mode reaching 100M U.S. visits faster than any Gen AI platform and shifting discovery into hybrid search–AI interfaces. At the same time, AI-native browsers like ChatGPT Atlas and Perplexity Comet are redefining how users interact with the web, turning the browser itself into an intelligent assistant.

Another key shift is the volatility of AI citations. Changes such as Google’s September algorithm update caused a sharp drop in Reddit citations, illustrating how quickly AI retrieval patterns can shift, and how directly they can affect brand visibility.

Finally, LLMs continue to prioritise deep, authoritative, and niche content, as well as trusted external sources like publishers and UGC platforms. Brands that invest in long-tail content and credible off-site signals will be best positioned as AI systems become more selective in what they surface.

Here is what one SEO market leader is saying:

“Generic, educational content no longer works for brands attempting to win on LLMs. Generative engines have completely eliminated the top of the funnel, as they simply provide the answers to queries without sending any referral traffic.

Truly winning on AI search demands tactical, bottom-of-the-funnel, niche content that answers the user’s query in the shortest amount of time in the most tactical way.

This is exactly why original data and building brand authority have become the two most important factors for SEO and GEO.

With so much content being produced at scale, generative engines have no other choice, than to refer traffic to the most authoritative source.” – Austin Heaton, SEO Consultant

Winning the next phase of AI discovery

Generative AI has already rewritten the pathways of online discovery, and its influence is accelerating. User behaviour is shifting, platforms are fragmenting, and the mechanics of visibility are being reshaped in real time. In a landscape this fluid, gut instinct isn’t enough. Brands need a data-driven understanding of where they stand, how they’re being surfaced in AI answers, and what signals engines are responding to. The winners will be the ones who track these shifts early and use real insight to stay ahead of the next wave.

FAQs

How is generative AI changing the way people search online?

Generative AI shifts discovery from keyword lookups to conversational queries, synthesised answers, and deeper intent-driven interactions. This changes not just how people search but where decisions are made.

Why are citations becoming more important in AI answers?

Because LLMs increasingly rely on live web data. Citations reveal which sources AI trusts, and which domains are shaping AI’s recommendations.

What makes AI-driven referral traffic different from search traffic?

AI referrals are smaller in volume but significantly higher in intent. Users who click through tend to spend more time on-site and convert at higher rates.

How can marketers measure success in the AI era?

Traditional rankings matter less. New key metrics include visibility in AI responses, citation frequency, sentiment in recommendations, and off-site influence.

Why is ongoing data essential for competing in AI search?

Because AI models update frequently, citation patterns shift, and platform behaviour changes rapidly. Up-to-date data is the only way to understand where you’re gaining (or losing) visibility in real time.

by Darrell Mordecai

Darrell creates SEO content for Similarweb, drawing on his deep understanding of SEO and Google patents.