Baidu Unveils Over a Dozen AI-Native Applications
-
On October 17, at the Baidu World 2023 conference, Robin Li, founder, chairman, and CEO of Baidu, showcased over a dozen AI-native applications, including a revamped search engine and maps, aiming to "expand everyone's imagination and inspire the creation of even more impressive AI-native applications."
According to Li, the birth of AI-native applications is attributed to the four core capabilities of large models: understanding, generation, logic, and memory. "These capabilities were absent in previous eras, unlocking infinite possibilities for innovation," he said.
At the conference, Li announced that Baidu's search engine, Ruiliu (a productivity platform), maps, cloud storage, and Wenku (a document-sharing platform) will all be presented in a completely new form. He emphasized that the purpose of sharing these applications is to expand imagination and encourage more people to develop groundbreaking AI-native applications.
Specifically, Li introduced Baidu's new search engine, which features three key characteristics: ultimate satisfaction, recommendation stimulation, and multi-round interaction. When users search for information, the new search engine will no longer provide a list of links but instead generate multimodal answers—including text, images, and dynamic charts—based on content understanding, allowing users to obtain answers in one step. For complex queries, the "multi-round interaction" feature can meet more personalized search needs through prompts and adjustments.
Additionally, Li demonstrated Baidu GBI, a generative business intelligence product built with AI-native thinking. Unlike traditional BI software, which has high barriers to entry and complex data analysis requirements, Baidu GBI enables users to perform data queries and analysis tasks through natural language interaction. It also supports the injection of professional knowledge to meet more complex and specialized analytical needs.
By leveraging the understanding and regeneration of massive documents, images, and videos, Baidu's cloud storage and Wenku platforms now possess creative capabilities. For instance, the cloud storage service can not only pinpoint specific frames in a video but also summarize an hour-long video in seconds, extracting key quotes and highlights. Wenku can draft articles and create PowerPoint presentations based on vast amounts of high-quality data.
Baidu Maps and the intelligent office platform Ruiliu have also evolved into more attentive travel guides and super assistants through capabilities like understanding and memory. With Baidu Maps, users can simply voice their needs, and the app will leverage thousands of service interfaces to recommend restaurants, compare multiple locations, and provide travel suggestions. Ruiliu addresses workplace pain points, such as information overload in group chats, by "highlighting key points in one second." Its travel assistant can not only book flights and hotels but also summarize background information and talking points for client meetings by integrating with corporate systems like CRM.
Li believes that large models will usher in a thriving ecosystem of AI-native applications. Among these, plugins—a special type of AI-native application—offer low barriers to entry and ease of use, enabling developers and entrepreneurs to quickly join the ecosystem.
It is reported that personal and enterprise data, capabilities, or applications can quickly be transformed into AI plugins to enhance the capabilities of large models, making them more practical and user-friendly. Robin Li stated that a month ago, Baidu launched the Lingjing Plugin Platform, which has already attracted 27,000 developers to apply for entry, covering fields such as law, workplace, and education.
When developing AI-native applications, the foundational capabilities of large models are crucial. Robin Li explained that APIs are the primary method for AI-native applications to call upon foundational large models. Enterprises and developers can access large model APIs on Baidu's Qianfan Large Model Platform. Currently, the platform hosts 42 mainstream large models, covering nearly 500 scenarios across various industries.
"China has a wealth of application scenarios, and users are naturally inclined to embrace new technologies. With advanced foundational large models, we can build a thriving AI ecosystem," said Robin Li.
Additionally, Robin Li mentioned that future AI-native applications will undoubtedly be multimodal, extending beyond the information world to reconstruct the physical world. Autonomous driving is a prime example of how visual large models can reshape the physical world. Large models will enable Baidu's autonomous driving capabilities to surpass experiential systems, handle complex scenarios more intelligently, and achieve broader spatiotemporal coverage. Currently, Baidu's autonomous ride-hailing service platform, Apollo Go, has provided over 4 million rides.