Baidu's He Junjie: AI-Native Applications Will Become the 'Springboard' for Innovative Growth
-
On November 28, at the 'WISE2023 Business King Conference,' He Junjie, Senior Vice President of Baidu Group and General Manager of Baidu Mobile Ecosystem Business Group, delivered a speech titled Where AI Flows, elaborating on Baidu Mobile Ecosystem's reflections and practices during this year of industry-wide transformation driven by large models. He Junjie stated that every enterprise and organization is now contemplating how to embrace this new era. 'For all companies, the key question is whether large models have reduced costs, increased revenue, and accelerated growth—this is the essence,' he noted. He believes that a flourishing ecosystem of AI-native applications will act as a 'springboard' for innovative growth, while large models will demonstrate their 'Moore's Law' effect in this era, ushering industries into a new phase of exponential expansion.
Since the explosive popularity of ChatGPT, hundreds of companies worldwide, large and small, have been training their own large models. In response, He Junjie remarked, 'In the AI-native era, we need a million-scale of AI-native applications, not a hundred large models.' Baidu Mobile Ecosystem serves as Baidu Group's primary testing ground and first stop for implementing AI-native applications in this transformative landscape.
Over the past six months, leveraging the evolving capabilities of ERNIE and the ERNIE large model, Baidu Mobile Ecosystem has comprehensively restructured its application ecosystem and services, creating a diverse matrix of AI-native applications. 'This is not integration or mere access—it’s a complete 'redo,'' He Junjie emphasized. Products such as the new ERNIE App, Baidu Search, Baidu Docs, Baidu Input Method, and Baidu Tieba, along with enterprise platforms like Baidu Qingge and Qingduo, now offer entirely new intelligent experiences.
Among these, Baidu Search, as the cornerstone of Baidu, has been the focal point of this restructuring. The revamped search experience features three key attributes: ultimate satisfaction, recommendation-driven engagement, and multi-turn interaction. When users pose a question, the new search no longer returns a list of links but generates multimodal answers—text, images, dynamic charts—based on content understanding, delivering instant satisfaction. He Junjie revealed that Baidu Search now handles over 300 million generative search queries daily.
Baidu Docs exemplifies the transformative power of AI-native thinking. It has颠覆ed decades-old office software habits, evolving from a 'content tool' into a 'productivity tool' with capabilities like AI-powered PPT generation, document creation, and smart editing. These innovations have attracted 30 million users, with new features used over 200 million times, significantly boosting paid subscriptions.
Recently, Baidu released the ERNIE 4.0 version, which is the most powerful iteration of the ERNIE large model to date. It shows significant improvements in four key capabilities: understanding, generation, logic, and memory. ERNIE 4.0 also further enhances Baidu's AI-native application capabilities, accelerating their practical implementation.
He Junjie believes that growth is currently happening closest to large models, and the flourishing application ecosystem driven by these models will usher in new development opportunities for industries transitioning from a "stock market" to an "incremental market" landscape. Baidu is continuously building a "1+N" large model ecosystem, expanding from the foundational "1" (the large model) to the diverse "N" (AI-native applications).
Taking Baidu's mobile ecosystem as an example, it encompasses a creator community of 6 million, a commercial client base of millions, and a developer plugin ecosystem. These users, clients, and partners are all benefiting from the rapidly evolving "AI-native application ecosystem."
Among these, the large model plugin platform "Lingjing Matrix" empowers developers by integrating Baidu's large model capabilities with its traffic pool. Currently, the plugin ecosystem has received registration applications from 27,000 developers, covering over 20 vertical industries.
According to Baidu's latest Q3 financial report, both revenue and net profit exceeded market expectations, with the mobile ecosystem performing steadily. Advertisers have responded positively to new AI features, with those adopting them achieving high single-digit conversion rate growth on average in Q3. During the earnings call, Robin Li, Baidu's founder, chairman, and CEO, stated that Baidu is reconstructing its advertising system using the ERNIE large model, including services like generative creativity and generative targeting. These initiatives are expected to generate hundreds of millions of yuan in incremental revenue in Q4. Essentially, the revamped AI-native applications will serve as a "springboard" for sustained innovative growth.
Moving forward, Baidu will dedicate more innovation efforts and resources to building a diverse application ecosystem. Together with the industry, it aims to advance large models from 1 to N, fostering a thriving AI ecosystem.