Google AI in searches: already default in 43% of queries

What happened?

According to a TechCrunch report based on recent BrightEdge data, Google's AI Overviews —AI-generated answers that appear at the top of search results— are now shown in 43% of queries. This marks a significant increase from previous months, when the figure hovered around 15-20% in May 2024, according to data from the same analytics tool. The company has gradually expanded this feature since its experimental launch in May 2024, initially limited to certain types of questions (such as informational queries) and regions (mainly the United States). In August 2024, Google announced expansion to six new countries, including India and Brazil, and to more languages. The jump to 43% reflects much faster adoption than expected, surpassing analyst projections of 30% by the end of 2024.

Why is this important?

Google processes over 8.5 billion searches per day globally (according to Internet Live Stats). That nearly half of them include an AI summary means millions of users get answers without needing to click any link. This disrupts the business model of publishers and websites that rely on referral traffic. It also changes the user experience: search is no longer a list of links but a direct conversation with AI. According to a Google study, users value immediacy but have also reported that information is sometimes less detailed than from a website. The impact is comparable to the introduction of featured snippets in 2014, but on a much larger scale, as AI now generates original content rather than extracting snippets.

Consequences for the digital ecosystem

Impact on web traffic

Previous studies by consultancy Similarweb indicate that AI Overviews can reduce click-through rate (CTR) by up to 20-30% on informational queries, especially those not requiring transactions. Media publishers like Forbes or The New York Times have already reported declines in visits from organic search of between 15% and 25% in categories such as health and finance, according to internal data leaked to The Verge. In the long term, this could force content creators to diversify their traffic sources (social media, newsletters, etc.) or adopt subscription models. For example, Vox Media has increased its investment in newsletters and podcasts in response.

Advertising and SEO

For advertisers, the visibility of ads at the top of the page could be affected if AI Overviews take up more space, although Google has stated that ads will continue to appear in prominent positions. According to a Search Engine Land analysis, click-through rates on search ads have decreased by 8% on queries with AI Overviews. As for SEO, traditional optimization strategies for featured snippets may become obsolete; now the goal is for AI to cite your content as a reliable source. Google has confirmed it prioritizes high-quality, authoritative content (E-E-A-T) for citations in AI Overviews, benefiting established sites but harming new ones.

Quality and misinformation

AI Overviews are not infallible. Cases of incorrect or absurd answers have been documented (e.g., suggesting eating a rock daily, as reported by The Guardian). Google claims to have safeguards, such as filters for harmful content and quality reviews, but the scale makes full human review impossible. A Stanford University study found that 12% of AI Overview responses contained false or misleading information on health topics. Reliance on AI for critical information (health, finance) poses risks, especially for vulnerable users. Google has implemented a feedback system to report errors, but transparency on how responses are generated remains limited.

What should readers know?

• Not everything AI says is true: always verify primary sources, especially on sensitive topics. For example, in medical queries, AI may cite non-peer-reviewed sources.
• Web traffic will change: if you are a content creator, prepare for less reliance on Google. Platforms like Substack or social media (TikTok, Instagram) are gaining relevance as traffic sources.
• Privacy matters: AI searches may collect more contextual data to personalize responses, raising questions about data use. Google says data is anonymized, but there are no independent audits.
• Alternatives exist: engines like DuckDuckGo or Brave Search offer experiences without integrated AI, though with lower accuracy on certain queries. There are also options like Perplexity AI, which combines search with cited answer generation.

“Google is transforming search from a directory of links to an answer assistant. 43% is just the beginning; the goal is 100%.” — Analyst at TheVortiq

Historical context

Google is not the first to integrate AI into search. Microsoft's Bing Chat (now Copilot), based on GPT-4, launched in February 2023, but its market share is below 4%. However, Google's market share (over 90%) makes its move massively impactful. The rapid adoption of AI Overviews echoes the introduction of Google Discover (2017) or featured snippets (2014), but with a key difference: AI now generates new content, not just extracts snippets. Additionally, the regulatory context is different: the European Union is investigating whether AI Overviews violate the Digital Markets Act by favoring Google's own services. In 2023, the European Commission fined Google €4.3 billion for anti-competitive practices in Android.

Looking ahead

Google is likely to continue increasing the percentage of searches with AI Overviews until it becomes universal, possibly by 2025. It may also integrate multimodal responses (images, videos) and personalization based on history, as it already does with Gemini. According to Sundar Pichai, Google's CEO, the vision is for search to be a “proactive assistant” that anticipates needs. The open question is whether users will passively accept these answers or seek alternatives. For now, convenience seems to win: in a Pew Research survey, 68% of US users said they were satisfied with AI-generated search answers. However, 45% expressed concern about accuracy. The balance between speed and reliability will define the future of search.