Baidu CIO Li Ying: Large models are an important opportunity in the enterprise office field, and the native reconstruction of AI will change the way intelligent work is done.

On August 16, 2023, the WAVE SUMMIT Deep Learning Developer Conference was held in China. The event was hosted by the National Engineering Research Center for Deep Learning Technology and Applications and hosted by Baidu Feipiao and Wenxin Big Model. At the meeting, Baidu released the latest progress and ecological achievements of a series of technologies, products, such as Wenxin large model, flying paddle platform and AI native applications such as Liuliu. Baidu Group Vice President and Chief Information Officer Li Ying delivered a keynote speech. She believed that the current fourth technological revolution with AI large models as the core technology will fundamentally promote productivity changes, provide strong support for all walks of life, and provide enterprises with The office field brings unprecedented development opportunities

Based on AI native thinking, Li Ying announced that Baidu’s smart work knowledge management concept “Innovation Pipeline = AI Pound upgrade. Li Ying mentioned that large models compress human understanding of the world, and intelligence emerges. The most important manifestation is excellence in knowledge. Large models are a major opportunity in the corporate office space and can lead to more profound changes

Li Ying pointed out that the big language model, represented by Wen Xinyiyan, already has the capabilities of understanding, generation, logic, and memory, and has led to the emergence of more capabilities, bringing major changes to intelligent work. This change has three main characteristics: first, the interaction method will be mainly based on natural language interaction; secondly, the demand satisfaction method can achieve end-to-end ultimate satisfaction; finally, the work model and process will be reshaped, and ultimately based on AI native construction A new working paradigm

Ruliu "Super Assistant" is officially released, demonstrating the native AI capabilities of four major scenarios on site

At the meeting, Li Ying released an important product called "Super Assistant", which is based on Wen Xinyiyan's ability and has been fully upgraded on the basis of the original Ruliu Super Assistant. Different from the assistant concept, this product is more humane and proactive

The goal of Ruliu Super Assistant is to create an assistant for everyone that can understand you, be professional, and accompany you in your work in real time. Li Ying focused on demonstrating the functions of the super assistant in four scenarios including mobile task execution, intelligent document processing, conversational business intelligence CBI and intelligent communication, and conducted more than 10 demonstrations

As the entrance to AI capabilities, the super assistant can be evoked and respond anytime and anywhere, accompanying the user at all times. The super assistant’s voice functions such as one-click appointment, one-click vacation, and one-click travel are demonstrated on mobile devices. For example, the user can voice input on the super assistant interface "I will go to the Shanghai R&D Center for a meeting the day after tomorrow morning and return to Beijing on the same day. Please help me plan my itinerary." The super assistant can quickly call the functions of different modules or platforms and automatically submit travel applications and reservations. Air tickets and hotels, and provide ideal accommodation locations based on user habits. It can also intelligently complete corporate settlements without requiring users to go through the reimbursement process, just like a real assistant. Operation efficiency is compressed from hours to minutes or even seconds.

In intelligent document processing scenarios, the super assistant has excellent capabilities and can help users find documents, learn knowledge and read literature. Just enter the scene in which the document appears, and the super assistant can accurately find the target document. For English papers, the super assistant can span different platforms, quickly generate Chinese abstracts with one click, and has translation and interpretation functions. Super assistants not only greatly improve the efficiency of knowledge acquisition, but also greatly expand the boundaries of employees' capabilities

Super Assistant has also developed conversational business intelligence based on large model technology, namely Baidu Conversational BI (CBI). In the "Baidu CBI" scenario, Li Ying demonstrated functions such as querying sales data, analyzing team data, and querying business opportunities. Different from traditional business intelligence, through the super assistant, structured business analysis results can be quickly calculated and output using natural language input, and CBI will even provide suggestions

Communication is one of the most common scenarios at work, and super assistants can help reduce noise and burden, extract key information and focus on the core when faced with a large amount of information. Super Assistant has multiple functions in communication scenarios, such as "Super Assistant IM Intelligent Summary", which can generate intelligent summaries after selecting multiple conversations and forward them to the recipient with one click. The recipient can quickly see the concise and accurate summary, thus Make decisions and respond quickly. According to statistics, this feature increases message reading efficiency by 3 times. In terms of meetings, AI insights and AI meeting minutes can improve the reading and application efficiency of meeting content by 3.5 times

The Comate series of products has been significantly upgraded, and Comate X is open to enterprise developers for the first time

Baidu Comate series products have undergone a new upgrade in the development of intelligent scenarios, and have released intelligent programming assistants Comate X and Comate Stack tool suites

Comate In addition to being able to interpret code and generate code, it also has the ability to generate interline comments, unit tests, documentation, command line and interface code. Currently, the assistant supports more than 30 programming languages ​​​​and more than ten integrated development environments

At the scene, Li Ying cooperated with Baidu engineers and used natural language to propose a task to Comate X, which was to "write an activation code program." Surprisingly, within a few minutes, the entire process of code generation, code interpretation, code annotation, and unit testing was completed. More importantly, during the entire process, the engineer did not write a line of code, but relied entirely on the programming assistant to complete

Li Ying officially announced that Comate X is open to enterprise developers, becoming the first commercial full-scenario intelligent programming assistant in China. Baidu has widely used Comate series products internally, and has currently provided services and docking with more than 100 corporate partners, entering a mature commercialization stage

Comate X is a R&D tool chain that uses AI native reconstruction. At the same time, this tool chain also provides services for the development of AI native applications. The research and development process is also undergoing reconstruction, requiring a new architecture and tool suite adapted to AI applications. Based on this, Li Ying released the Comate Stack tool suite, which can support the entire process of AI native research and development. Using Baidu Comate Stack, you can develop and launch a super assistant plug-in in just two steps. Through the Comate Stack tool suite, the development threshold for AI native applications can be significantly lowered and the efficiency of AI native transformation can be improved

Baidu Intelligent Work Platform has currently empowered dozens of major industries and fields, including finance, energy, artificial intelligence, manufacturing, communications, etc. Among them, the Comate series of intelligent products developed have been widely used in all walks of life

The corporate office field is undergoing a major change. This change is accelerated by artificial intelligence represented by large language models, which accelerates the transformation and upgrading of all walks of life. In this process, technological innovation and application implementation have formed a virtuous cycle, continuously improving capabilities such as understanding, generation, logic, and memory, while also continuously expanding the breadth and depth of industrial applications. Large language models bring glimmers of hope for general artificial intelligence

The above is the detailed content of Baidu CIO Li Ying: Large models are an important opportunity in the enterprise office field, and the native reconstruction of AI will change the way intelligent work is done.. For more information, please follow other related articles on the PHP Chinese website!