(Yicai) Aug. 24 -- An increasing number of consumers in China, especially the younger generation, use artificial intelligence as their primary information and evaluation tool before buying products, bringing new challenges to manufacturers and brand managers, according to a new report.

Almost 80 percent of Chinese consumers ask AI tools such as ByteDance's Doubao and DeepSeek to help with purchase decisions, according to a second-quarter survey on consumer willingness released by the China Association of Small and Medium Commercial Enterprises and other institutions. AI has reshaped the entire process of buying goods, from collecting information to comparing prices, selecting brands, and evaluating purchases.

Active users on Doubao, Alibaba Group Holding's Qwen, and DeepSeek reached 382 million, 167 million, and 130 million, respectively, as of June 30, according to data from QuestMobile.

"The new generation has grown accustomed to using AI tools to understand a brand," Shen Yang, a professor at Tsinghua University's School of Journalism and Communication and School of AI, said at a recent seminar hosted by Fudan University's Center for Communication and State Governance Research in Shanghai.

The invisible new intermediary is rewriting the brand communication system that has been effective for decades, shifting it from "being seen by users" to "being selected by AI" to keep up with most consumers' new habit of "asking AI before placing an order."

"The main battlefield influencing brand reputation is transforming from public web pages to how brands are perceived and represented by AI systems," said Li Zhengrong, co-president of Midu Technology.

How a company monitors and optimizes the AI's perception of its products is a "headache," noted Huang Xiaojun, chief executive of AI-native enterprise Yichen Technology.

"In the AI era, profound changes are taking place in the underlying links of information dissemination, with brand management facing new structural challenges," said Zhang Zhi'an, director of Fudan University's Center for Communication and State Governance Research.

In an era where AI tools have become vital for consumer decision-making, brand images are being fragmented into countless versions, according to a report on brand health and management innovation in the AI era, led by Zhang, released at the seminar. Descriptions of a brand can differ drastically across AI models, over time, and depending on conversational contexts, while companies have no control over what data AI is trained on, which sources it cites, or what wording it uses to describe them.

In addition, brand management faces the risks of external forgery and cognitive manipulation. Malicious competitors need only spend a few dozen or a hundred yuan (about USD10) to flood the internet with false information, contaminating AI models' understanding of a brand within hours. Once such negative information enters the model's knowledge base or searchable scope, it will persist for a long time.

In the past, brands competed for visible traffic exposure, but competition has shifted to the cognitive level in the AI era, Zhang said. Although traditional platforms remain important,  it is far more crucial whether large language model knowledge bases can incorporate and credibly cite information from the brand itself, as well as the stance and phrasing these models subsequently use, Zhang added.

While brands previously fought to "be seen by users," the first hurdle they must now clear is to "be selected by AI," the report pointed out.

Brand competition is shifting from a short-term, bidding-style "traffic game" to long-term, structured "cognitive asset governance," the report added. The core of these cognitive assets is the knowledge system through which a brand is recognized, cited, and trusted by AI models, which directly determines whether a company is "selected by AI." 

"When identifying and citing content, AI focuses more on structural clarity, logical consistency, and data verifiability," Zhang noted. "It prioritizes highly reliable sources and synthesizes various materials to construct a corporate profile. This poses entirely new demands on companies' cognitive asset management."

"The cognitive approach to managing brand reputation is changing, requiring a shift from managing event outcomes to managing the entire process," Li stressed. Brand management is no longer a series of isolated crisis-response actions, but rather a long-term capacity-building effort spanning a company's entire lifecycle, Li added.

Brand management in the AI era is highly challenging, Zhang said, noting that the earlier a company explores how AI retrieves its brand information and optimizes its visibility across different AI platforms, the better it can enhance its brand image.

Brands in the AI era will inevitably be multimodal, involving text, images, video, and audio, Shen pointed out. The true essence of brand management in the AI era is transforming from controlling a single standardized narrative to designing a self-evolving, multi-version communication system, he said.

Editors: Tang Shihua, Martin Kadiev