Baidu Integrates Generative AI into Search App
Baidu has integrated its OpenClaw AI model directly into its main search app, reaching 700 million users. The update allows users to perform complex tasks like scheduling and code-writing within search, foreshadowing a similar convergence of Google Search and its Gemini models.
Baidu's AI chatbot, known as Ernie Bot, has surpassed 200 million monthly active users. The AI is integrated directly into the main Baidu search app, which itself boasts approximately 704 million monthly active users. This integration strategy aims to make AI functions readily available within an existing high-traffic product, rather than requiring users to download a separate application. The AI, built on Baidu's "Enhanced Representation through kNowledge IntEgration" (ERNIE) framework, is designed to understand the nuances of the Chinese language and culture. This allows it to perform tasks beyond simple search, including booking flight tickets, ordering food delivery, and offering health and legal advice through integrations with platforms like JD.com, Meituan, and Trip.com. The latest version, ERNIE 4.0, is claimed by Baidu to have capabilities rivaling those of OpenAI's GPT-4. This move by Baidu is part of a larger trend and escalating AI competition among Chinese tech giants. Alibaba has embedded its Qwen chatbot into its consumer apps, and ByteDance has rolled out an AI-integrated phone. Baidu's CEO, Robin Li, has emphasized that in the AI era, applications are paramount, and the company is focusing on "Search agents" as a primary area of development. The strategy mirrors Google's integration of its Gemini AI model into its search products, known as the Search Generative Experience (SGE). Both companies are moving away from a simple list of links, instead providing AI-generated summaries and conversational abilities directly within the search interface. By embedding these powerful AI models into their core, high-frequency products, both Baidu and Google aim to gradually shift user behavior and lower the barrier to adopting generative AI features.
