Source | Bohu Finance and Economics (bohuFN)

Recently, Kimi quietly launched its exploratory version, which even briefly trended on social media. Moon's Dark Side remarked that the Kimi exploratory version was the team's hard work during the National Day holidays, but they didn't anticipate the overwhelming response from users.

It is reported that the Kimi exploratory version possesses AI-driven autonomous search capabilities, simulating human reasoning processes to decompose complex problems into multiple levels, conduct deep searches, and promptly reflect on and improve results, thereby assisting users in efficiently completing complex tasks such as analysis and research.

However, Kimi is not alone in focusing on AI search. Not long ago, Baidu announced the upgrade of Wenxin One to "Wenxiaoyan," positioning it as an intelligent assistant for "new search."

With the support of large models, the traditional search industry has frequently sprouted "new flowers," including native AI search tools like Perplexity and Meta, as well as AI chatbots like Kimi and Tiangong AI vying for market share. The search market is no longer solely dominated by Baidu.

While new players continue to explore search scenarios, Baidu, as an established player, is actively defending its turf. Does AI search stand a chance in transforming the traditional search industry?

01 Targeting search scenarios, Kimi "learns from others' strengths and compensates for its weaknesses"

After the official launch of the Kimi exploratory version, its product lead once stated, "If Kimi cannot find the information, it is highly likely that users will also struggle to locate it through traditional search engines." Clearly, Kimi aims to revolutionize the traditional search paradigm.

In traditional search engines, represented by Baidu, users pose questions, and the engine crawls web pages to gather information, ranks it according to its own logic, and presents it to users.

There are two underlying pain points here. Firstly, users need to phrase their questions simply and directly, potentially breaking them down into multiple queries. Secondly, it's difficult to avoid the influence of advertising mechanisms in the ordering of search results.

The Kimi exploratory version attempts to address these pain points by decomposing complex queries into simpler steps, guiding the model to autonomously reason and solve problems, shifting the large model's approach from "fast thinking" to "slow thinking." Notably, OpenAI's recently released o1-preview model also adopts a "slow thinking mode."

In the past, large models, while "intelligent," did not truly think like humans, merely mimicking human behavior through extensive data training. However, under the "slow thinking" logic, future large models will genuinely possess autonomous thinking and reflection capabilities.

According to Moon's Dark Side's demonstration, users previously had to search for each company individually to find out how many of the top 50 tech companies had headquarters in the capital. In contrast, the Kimi exploratory version can swiftly scan web pages and summarize the answer. It is understood that the search volume of the Kimi exploratory version is ten times that of the regular version, enabling users to thoroughly analyze over 500 pages with a single search.

Beyond data integration, the Kimi exploratory version excels in in-depth search, research, and analysis. For instance, when asked which stock, BYD or Moutai, would yield a higher return if purchased during the same period, Kimi can directly calculate the results.

In this query, Kimi showcases its search and learning capabilities by decomposing the complex question, gathering relevant data for both stocks, learning how to calculate "return rates," and ultimately providing an answer.

Additionally, the Kimi exploratory version enables fuzzy searches. When asked, "Which Silicon Valley company has produced a product similar to the iPhone?" Kimi can infer the company name from key details and even self-reflect to supplement more information.

With the launch of the Kimi exploratory version, Moon's Dark Side has further demonstrated its intention to delve deeper into search scenarios. As early as July this year, Kimi introduced a browser plugin version offering highlighting, questioning, and summary features to enhance users' search efficiency.

In recent years, the domestic and international search markets have undergone significant changes, with new players such as large model giants, native AI search tools, and AI chatbots entering the AI search arena. OpenAI, a major large model player, also released its AI search product, SearchGPT, in July this year, presenting an opportunity for Kimi, which is still seeking a commercial path.

On one hand, search scenarios can further leverage Kimi's strengths. Considering that Kimi's core user base primarily comprises knowledge workers and university students with high-frequency search needs, search aligns perfectly with AI chatbot functionality, enabling quick responses and clear links to sources.

On the other hand, traditional search engines are gradually declining in popularity as young people turn to specialized platforms for different search tasks, such as "Dianping" for restaurant recommendations and "Xiaohongshu" for travel planning, presenting an opportunity for Kimi.

However, the Kimi exploratory version is not without its limitations. Firstly, large models still struggle with "hallucinations," sometimes sourcing data from other users' articles, which may compromise data credibility and timeliness.

Furthermore, some users report that when queries are vague, the Kimi exploratory version struggles to comprehend user intent, with overall reasoning abilities still inferior to OpenAI's o1-preview model.

02 Focusing on "new search," Baidu wages a defensive battle

Nevertheless, Kimi's imperfections do not hinder its challenge to Baidu, given users' long-standing frustration with the latter in the search market.

As a gateway to information, search engines have always been a highly contested territory among internet companies, as controlling a search engine equates to wielding influence over information dissemination and connecting various service ecosystems.

In the early days of the internet, Baidu, as a pioneer in the search market, leveraged its technological lead and content ecosystem to establish a dominant position in Chinese-language search. Despite competition from other search engines, Baidu maintained its "search monopoly" status due to its brand recognition and adherence to policies and regulations.

However, with a single search entry point, users began to feel constrained, particularly criticizing Baidu's "paid ranking" mechanism, where advertisers could pay to have their content prominently displayed, potentially compromising information authenticity.

According to Baidu's 2005 prospectus, its online promotion revenue once accounted for 90% of its total income, but this trend is changing. From the second quarter of 2023 onwards, Baidu's online marketing revenue grew by 15%, 5%, 6%, and 3%, respectively. In the second quarter of this year, Baidu's online marketing revenue was 19.2 billion yuan, a year-on-year decrease of 2%, marking the first negative growth.

Baidu's slowing ad revenue can be attributed to two factors. Firstly, emerging social media platforms like short videos and live streams have eroded the share of traditional search ads in the overall internet ad market. Secondly, users are increasingly diverting to specialized platforms for search tasks, with AI search gradually replacing traditional search and eroding Baidu's market share.

According to the latest Statista data, while Baidu remains the top search engine in China with a 55.85% market share in May 2024, this is a significant drop from the 86.82% share it held in November 2021.

Naturally, Baidu is aware of the market challenges. Early in the large model boom, Baidu's Chairman Robin Li stated that all products must be rebuilt using large models, with search services poised to become the "killer app" of the AI era.

Baidu has also attempted to reinvigorate its search business with AI. In May this year, Baidu management revealed that approximately 11% of search results were generated using AI technology, with this proportion expected to increase in the future.

However, despite Baidu's early start in large models and its exploration of AI search possibilities years ago, it has yet to deliver significant surprises.

In May this year, Baidu launched an AI smart Q&A feature that acts as an intelligent assistant when users search for information, refining web content and summarizing answers. However, compared to other AI search tools, Baidu's responses lack depth and breadth, struggling with questions requiring reasoning and inference.

Nevertheless, a month ago, Baidu upgraded Wenxin One to "Wenxiaoyan," positioning it as a "new search" intelligent assistant, unlike Kimi, which is further enhancing its search capabilities. Started as a search engine, Baidu chose to introduce a smart assistant focused on "new search."

Currently, the primary differences between "Wenxiaoyan" and other AI search products lie in two aspects. Firstly, it leverages Baidu's extensive ecosystem support. For instance, when users inquire about transportation routes or travel guides, "Wenxiaoyan" can directly access Baidu Maps to provide comprehensive lifestyle services such as transportation and dining options.

Secondly, "Wenxiaoyan" offers five core scenarios: asking questions, chatting, writing articles, drawing images, and assigning tasks. Beyond AI search, Baidu provides users with a wide range of personalized AI services, essentially offering a comprehensive "AI suite."

While both Baidu, the "veteran" in the search industry, and Kimi, the "newcomer," are vying for market share, their strategies differ. The former pursues diversification, while the latter focuses on specialization. However, both share a common goal: addressing their respective ecosystem shortcomings.

03 Assessing the odds for "Kimi-like" entrants invading Baidu's stronghold

In the long run, AI search represents a major trend, attracting numerous cross-industry players seeking a piece of the pie. As a traditional player, Baidu is actively integrating AI capabilities and exploring new product models.

However, transforming the traditional search industry with AI search is far from straightforward. Firstly, while large models lower the entry barrier for search products, most AI search offerings follow a simplified search path and enhanced information matching approach, resulting in AI engines appearing convenient but sometimes devoid of substance.

Moreover, users opt for AI search products hoping to simplify complex searches. However, given issues like information source pollution and large model "hallucinations," users may still need to re-verify critical information, highlighting the challenges in building absolute trust in AI search.

Secondly, commercialization poses a dilemma. Both Baidu and Kimi face pressure to monetize their search platforms. Kimi remains in the investment phase, and most large models have only two primary monetization paths: TOB (with intense competition and thin margins) and TOC (reliant on subscription fees, requiring outstanding product capabilities to cultivate user willingness to pay), which is why Kimi continues to refine its search functionality.

Additionally, industry insiders reveal that AI search is just the first capability of the Kimi exploratory version, with more to come, possibly including multimodal abilities. This suggests that like Baidu, Kimi aims to continuously enrich its "AI suite."

Baidu's stance is more ambivalent. From a commercial perspective, AI search significantly disrupts traditional search models by directly providing answers, making it challenging to integrate ads. Consequently, Baidu's once lucrative search business now faces the need for reinvention.

Thus, until these two issues are thoroughly resolved, AI search is unlikely to fully replace traditional search engines. Nevertheless, enhancing search efficiency and accuracy through large models is already reshaping the search market landscape.

Currently, "Kimi-like" entrants are attempting to penetrate Baidu's stronghold, but Baidu won't sit idle. The evolution of search engines inherently involves industry innovation, with each technological breakthrough and player entry/exit potentially altering the search market landscape.

Before these challenges are addressed, AI search products must actively explore product and monetization models, striking a balance between commercialization and technology while enriching content ecosystems to deliver more precise and personalized search results, thereby effectively attracting and retaining users.

In this transformative battle for search engines, it's uncertain who will emerge victorious. However, one thing is clear: the search market will no longer be solely dominated by Baidu.

The cover image and accompanying illustrations belong to their respective copyright holders. If the copyright owner believes their work is not suitable for public viewing or should not be used free of charge, please contact us promptly, and we will make the necessary corrections immediately.