A new report from SparkToro and Datos shows that artificial intelligence tools expanded quickly in the United States over the past two years but are now gaining users at a slower pace. About 38 percent of Americans use assistants such as ChatGPT, Claude, Copilot, Gemini, Perplexity, or Deepseek at least once a month, compared with 8 percent in early 2023. Around one in five rely on them heavily, meaning more than ten times each month.
The swift expansion that defined 2023 and early 2024 has eased. Since September 2024, adoption has risen only gradually, suggesting that most of the early demand has already been met. Unless developers reach new audiences outside professional and academic fields, usage may be nearing a plateau.
Traditional Search Remains Stable
Despite greater awareness of AI assistants, search engines remain dominant. The study found that 95 percent of Americans use Google, Bing, Yahoo, or DuckDuckGo monthly. Nearly nine in ten qualify as heavy users, a share that has grown since 2023. The average number of searches per person has also edged upward, reinforcing the sector’s steady position.
Seasonal patterns are visible in the data. For example, activity dipped in June 2025 when schools closed for summer, reflecting the fact that many queries, especially through AI, come from education-related use. This drop mirrored the decline in OpenAI’s token usage during the same period.
AI and Search Work Together
The findings counter the claim that AI assistants are drawing people away from search engines. Instead, the evidence shows that users of AI tools often perform more searches, not fewer. A separate study by Semrush linked ChatGPT adoption to an increase in Google search activity.
Defining the boundary between AI and search is also becoming more difficult. Google has integrated Gemini into search results with features such as AI Overviews. Bing relies on Copilot and OpenAI models, while DuckDuckGo uses Bing results with its own AI enhancements. As a result, most search services now operate as hybrids of indexing and machine learning.
Industry Response and Criticism
Reactions to the report varied. Some observers argued that calling Google or Bing “traditional search” underplays the role of AI in their products. Others questioned the accuracy of the usage measures, noting that changes in how platforms route traffic can affect results.
Critics also drew a line between adoption and impact. ChatGPT may reach hundreds of millions of people daily, yet that has not produced a measurable fall in search engine reliance. Analysts compared the pattern to television streaming, where households often add services rather than replace them.
Marketers paid closer attention to how behavior is shifting. They noted that while users may visit fewer websites, visits tend to carry more intent. Since AI assistants often return similar answers, businesses will need clearer strategies to stand out when fewer but more targeted visits occur.
Methodology and Context
The report relied on Datos’ clickstream data from millions of U.S. desktop devices. This approach avoids problems common in self-reported surveys but leaves out mobile apps, which are often where AI tools gain the most use. Separate figures from Europe and the United Kingdom place adoption about ten percentage points higher than in the U.S., though the overall pattern of slowing growth is similar.
Researchers also clarified that falling website traffic from Google does not mean search use is shrinking. Instead, more queries are being resolved within search results through snippets or AI-generated answers, reducing the need to click through to another site.
Conclusion
The data shows that AI assistants are widely used but that their early surge in growth has cooled. Search engines remain almost universal and continue to expand modestly. Rather than replacing search, AI appears to be developing as a complementary tool. For businesses, this means navigating a landscape where customer journeys involve fewer site visits, but those visits often carry greater weight.
Notes: This post was edited/created using GenAI tools. Image: DIW-Aigen.
Read next: Google Translate Adds AI-Powered Conversation and Language Learning Tools