Frost & Sullivan releases 'White Paper on China's AI Search Industry 2025': Focusing on Complex Scenarios for In-depth Intelligent Search Insights

AI search engines are gradually changing the landscape of the search engine industry, with traditional search engines seeing their market share declining year by year.

AI search is triggering a fundamental transformation in the search engine industry. The core difference between AI search and traditional search engines lies in the fact that AI search uses natural language processing (NLP) and machine learning technologies to deeply understand users' complex query intentions. It can directly integrate multiple sources of information to generate precise and concise summary answers presented to users. This has completely changed the traditional search model, which relied solely on keyword matching and returned a large number of web links for users to sift through on their own.

The market share of traditional search engines is facing a significant decline. For example, Baidu Search's market share has dropped from 86.8% in November 2021 to 55.9% in May 2024. The reasons for this change are not only due to the gradual diversification of the search engine market and the continuous changing user needs, but also the rise and popularization of AI search engines. Specifically, Baidu Search's AI intelligent visits have shown a significant increase, from 24,314 times in July 2023 to 39,199 times in July 2024.

User inquiry methods are shifting from keyword commands to structured sentence interactions.

The question-asking habits of AI search users have undergone a structural transformation, no longer limited to keyword stuffing as in traditional searches, but instead showing a strong preference for using structured sentences for interaction. More importantly, users actively incorporate specific details, specific needs, and application scenarios into their questions, providing rich contextual information to guide AI to generate more precise and personalized answers. This change reflects an increasing understanding of AI search tools among users and a learning to improve the quality of information acquisition through more efficient dialogue methods.

The reason why user questioning methods are becoming increasingly complex and mature is fundamentally that the core scenarios they use AI search for have shifted towards handling 'complex problems', especially overcoming the capacity bottlenecks of traditional search engines. It is precisely to solve these multi-dimensional tasks that require in-depth analysis and logical reasoning that users must construct more refined and targeted questions. Therefore, this new questioning paradigm clearly reveals that the market demand for AI search is developing in an increasingly vertical and long-tail direction, placing higher demands on AI tools' semantic understanding and integration of professional knowledge.

Application scenario differentiation: AI search mainly targets in-depth and complex problems rather than daily quick queries.

When users use AI search, their application scenarios have shown a clear differentiation trend. Most users tend to view AI search products as professional tools for solving complex problems, conducting in-depth analysis, and addressing traditional search bottlenecks, rather than their first choice for daily basic information queries. Users also tend to ask questions in complete sentences with clear structure, containing specific details and scenarios, rather than simple keywords, reflecting their demand for verticalization and long-tail characteristics.

Surveys show that although AI search products have a high average daily usage frequency, 76.6% of users still use them in conjunction with traditional search engines. This indicates that at this stage, AI search is more of a powerful auxiliary or enhancement tool, used to handle deep tasks that traditional search is not capable of handling, and has not yet completely replaced the role of traditional search in fast and convenient query scenarios.

User trust is the core bottleneck, and information traceability reliability becomes crucial.

Despite the increasing adoption rate of AI search, users still have a weak trust in its generated content, which constitutes the core bottleneck of its development. Research data shows that up to 90% of users will double-check the answers provided by AI search, and at the same time, 87.4% of users pay special attention to the source information of the answers. This clearly reveals the authority, transparency, and quality of information sources, which are the primary determinants affecting users' final acceptance of AI responses.

When conducting information traceability, AI search engines consider multiple dimensions such as semantic relevance, data source authority (such as government websites, academic journals), citation frequency, content timeliness, and source diversity to allocate weights, ensuring the reliability of results. However, current AI searches mainly rely on web crawlers to obtain data, but only about 10% of the data is of high quality, which is still concentrated in sources such as books, Arxiv, and encyclopedias. Therefore, improving the accuracy of answers and the reliability of traceability are urgent issues that all AI search products need to solve.