On the morning of November 9th, the 2023 World Internet Conference Wuzhen Summit "Internet Entrepreneurs Forum" was held. Alibaba Group CEO Wu Yongming delivered a keynote speech. Wu Yongming said that AI technology will fundamentally change the way knowledge iteration and social collaboration are driven, and the development acceleration driven by this will far exceed our imagination. At present, we are at a node where traditional computing and AI computing are switching importance levels, and AI computing will eventually take over traditional computing resources. The deep integration of AI and cloud computing will become an important driving force for cloud computing iteration.
The following is a transcript of Wu Yongming's speech, with slight deletions.
We see that the technological revolution triggered by this round of AI is a historic opportunity for the next stage of digital economy development. I made an observation on the main forum yesterday: the paradigm of computing is undergoing fundamental changes, and the importance of AI computing is surpassing traditional computing. AI computing driven by large models will ultimately become the cornerstone of the entire digital world, and become the scheduler and central nervous system of computing resources throughout the digital world, ultimately taking over traditional computing resources dominated by CPUs. Today, I would like to share a few personal observations and reflections along this basic viewpoint.
The first point to consider: AI technology will fundamentally change the way knowledge iteration and social collaboration are carried out.
Compared to previous digital technologies, AI will bring about revolutionary changes in information, changing the way knowledge iteration and social collaboration are carried out. Looking back at human history, it was only after the invention of language, writing, and printing that humanity achieved large-scale knowledge iteration and social collaboration across space and generations, gradually accelerating the pace of technological development - just like computer languages and hard disk storage in the information age.
Now, AI technology represented by large models enables machines to understand human language and logic, and has become increasingly powerful in intelligence. Machines carry enormous computing power, helping humans manage knowledge and facilitate collaboration, and the development acceleration driven by this will far exceed our imagination. Whether it is the dimension of productivity or the dimension of production relations; Whether in the digital world or the real world, AI technology will have a profound impact.
The second point to think about: We are currently at a node where traditional computing and AI computing switch important levels.
From the trend of rapid development of AI technology, AI computing will soon replace traditional computing as the cornerstone of the digital world. Nowadays, traditional CPU based computing must rely on the algorithmic logic preset by product managers and programmers, who translate real-world data flows into machine language to help people improve efficiency, to some extent similar to the human cerebellum. And when our machine can fully understand human language and the connections between real-world things through the new generation of AI technology, it has intelligence similar to the human brain.
In traditional software mode, many cases still rely on traditional calculations to call certain intelligent functions of AI. With the rapid development of AI technology, the future software model will become centered around AI big model driven computing, and AI computing will take over the traditional computing resources managed by programmers under the traditional computing mode.
The third point to consider: AI will directly understand user needs and enable software to have the ability to upgrade independently.
Because AI is increasingly able to understand human language and the real world, software can autonomously optimize and upgrade in the future, intelligently create new applications to meet the ever-changing needs of individuals and businesses.
Under the traditional software model, if a company's sales or financial policies change, product managers and programmers must first understand the requirements, modify the code, and launch the system. In the era of AI computing, some new phenomena will soon occur: assuming that a company's business system architecture is built on collaborative software with AI capabilities, such as DingTalk, the company's AI assistant may automatically upgrade the software and notify all employees at the same time as the company's sales director holds a decision-making meeting. It is even possible that historical data has been recalculated, and sales commissions have been recalculated (according to the latest policy).
We believe that such an era will soon come, where AI assistants will bypass programmers, directly understand users' needs, and automatically upgrade or create systems to meet their ever-changing needs. This is a visible trend.
The fourth consideration: The key to developing AI is a strong infrastructure and an open ecosystem.
The infrastructure that supports AI must be an infrastructure that integrates AI capabilities. The deep integration of AI and cloud computing will also become an important driving force for cloud computing iteration. We believe that the entire operation mode of user demand definition, application development, operation and maintenance management, and resource scheduling in the future will also be provided by cloud computing driven by large models. The dual wheel drive of "AI+cloud computing" is the underlying capability of Alibaba Cloud to face the future and support AI infrastructure.
The technological progress and industrialization of AI demand exponential growth in computing power, requiring a complete set of basic service systems in the AI era. Only through the systematic integration of advanced cloud computing and AI can we support the development and application of AI models throughout society. Alibaba Cloud is focusing on AI to comprehensively enhance the systematic basic capabilities of cloud computing. Currently, half of the country's major models are running on Alibaba Cloud, and we hope to lay a more solid foundation for our foundation.
The development of AI is neither desirable nor feasible to rely solely on a few enterprises, and it requires stimulating innovation and creativity throughout society. What Alibaba wants to do is to create a technology innovation platform for the AI era, serve enterprises and developers on the platform well, and work together to create a prosperous AI ecosystem.
To prosper the ecology, it is necessary to open up. Alibaba Cloud is one of the earliest enterprises in China to open source self-developed large models. It has successively opened up the 7 billion parameter model and 14 billion parameter model of "Tongyi Qianwen", and will also open up the 72 billion parameter model in the future. Many open source communities both domestically and internationally have developed industry applications based on our "Tongyi Qianwen".
Last November, we created the AI open source community "Magic Tower", which has attracted many enterprises and innovators since its operation for a year. At present, the Magic Building community has accumulated over 2300 AI models, attracting 2.8 million AI developers, and the model download volume has exceeded 100 million. We provide 10 hours of free GPU computing power for every AI developer in the magic building community.
Dear colleagues, the wave of AI technology revolution has just begun, and this is an important opportunity we are facing together. Alibaba will firmly increase its technological investment, continuously improve its basic service system capabilities, continuously expand its open borders, play the role of technology platform enterprises, and provide better services and support for industry development.