Notice of the General Office of the People’s Government of Beijing Municipality on Printing and Distributing “Several Measures to Promote the Innovative Development of General Artificial Intelligence in Beijing”

Notice from Beijing Municipality's General Office on "Measures to Promote General Artificial Intelligence Innovation in Beijing"

Source: Official Website of the People’s Government of Beijing

JPB Office [2023] No. 15

To all district people’s governments, municipal government committees, offices, bureaus, and municipal agencies:

With the approval of the municipal government, the “Several Measures to Promote the Innovative Development of General Artificial Intelligence in Beijing” are now issued to you for implementation in light of actual circumstances.

• Will the cryptocurrency policy in the US and Canada see a new direction in the 2024 presidential election?
• Interpretation of the "2020 Cryptocurrency Act": New Ideas for the Cryptocurrency Regulation Model
• Investigation | After the ban is lifted, Indian investors are more positive about cryptocurrencies

Beijing Municipal People’s Government Office

May 23, 2023

(This document is publicly available)

Several Measures to Promote the Innovative Development of General Artificial Intelligence in Beijing

In order to implement the Implementation Plan (2023-2025) for Accelerating the Construction of a Globally Influential Innovation Center for Artificial Intelligence in Beijing, and to give full play to the guiding role of the government and the catalytic role of innovation platforms,

I. Enhancing the Coordination and Supply Capacity of Computing Resources

(I) Organizing Commercial Computing Resources to Meet Urgent Needs

Focusing on the advantage of computing resources in this city, the computing partner plan is implemented, and existing computing resources are collected through strengthened cooperation with cloud service providers. The supply technical standards, hardware and software service requirements, computing supply scale, and supporting measures are clearly defined to provide diversified, high-quality and inclusive computing resources for innovation entities and to ensure that computing power needs for artificial intelligence technology innovation and product research and development are met.

(II) Efficiently Promoting the Construction of New Computing Infrastructure

New computing infrastructure construction projects are included in the computing partner plan to accelerate the construction of the Beijing AI Public Computing Center and Beijing Digital Economy Computing Center in the Haidian and Chaoyang districts, form large-scale and advanced computing supply capacity, and support the research and development of large-scale language models, large-scale visual models, multimodal large models, scientific computing large models, large-scale fine-grained neural network simulation models, brain-inspired neural networks and other technologies with parameters in the billions.

(III) Building a Unified Multicloud Computing Resource Scheduling Platform

For elastic computing resource requirements, a multicloud computing resource scheduling platform is built to achieve unified management and operation of heterogeneous computing environments, facilitate innovative entities to run various artificial intelligence computing tasks seamlessly, economically, and efficiently on different cloud environments. Further optimize the basic optical transmission network for direct connection of the regional computing clusters in Beijing with those in Tianjin, Hebei, Shanxi, Inner Mongolia and other areas, and improve the integrated scheduling capability of computing power in the Beijing-Tianjin-Hebei region.

Section 2: Enhancing the Supply Capability of High-Quality Data Elements

(4) Collecting High-Quality Basic Training Data Sets

Organize relevant organizations to integrate and clean Chinese pre-training data, form a safe and compliant open basic training data set; continue to expand multi-modal data sources, build high-quality text, image, audio, video, and other large-model pre-training corpora, and support data circulation and transactions in lawfully established data trading agencies.

(5) Planning to Build Data Training Bases

Accelerate the construction of a demonstration zone for data fundamental system pilot programs, explore the establishment of data training bases, promote the high-level opening of data elements, and improve the scale and quality of the city’s AI data annotation library. Encourage internet platforms providing high-quality corpus data for content information services for innovative entities to apply. Explore commercial cooperation scenarios based on data contributions and model applications.

(6) Building a Fine-Grained Annotation Crowdsourcing Service Platform for Data Sets

Build a fine-grained annotation platform for data sets in a crowdsourcing service manner, develop an intelligent cloud service system, and integrate relevant tool applications. Encourage and organize professionals from different disciplines to participate in the annotation of multi-modal training data and instruction data, and improve the quality of data sets. Research the incentive mechanism of the platform to promote its sustained and benign development.

Section 3: Building a General Artificial Intelligence Technology System

(7) Conducting Innovative Algorithm and Key Technology Research for Large Models

Around model construction, training, alignment, inference deployment, etc., actively explore innovative basic model architectures, study efficient parallel training technologies for large models and adjustment methods such as cognitive inference, instruction learning, and human intent alignment, and research efficient compression and end-side deployment technologies that support the inference of models with hundreds of billions of parameters, forming a complete and efficient technical system, and encouraging the construction of an open-source technology ecosystem.

(8) Strengthening the Research and Development of Large Model Training Data Collection and Governance Tools

Around the links of “collection, storage, management, research, and use” of training data, research full-amount real-time update technologies for internet data, integration and classification methods for multi-source heterogeneous data, construct relevant system of data management platform, and research algorithms and tools such as data cleaning, annotation, classification, annotation, and content review.

(9) Build a large-scale model evaluation open service platform

Encourage non-profit third-party organizations to build multimodal and multidimensional basic model evaluation benchmarks and evaluation methods; research model evaluation algorithms assisted by artificial intelligence, develop multidimensional basic model evaluation toolkits including generality, efficiency, intelligence, and robustness; build a large-scale model evaluation open service platform, establish a fair and efficient adaptive evaluation system, and automatically adaptively evaluate large models according to different goals and tasks.

(10) Build a large-scale model basic hardware and software system

Support the development of a distributed training system for large models to achieve efficient and automatic parallel training tasks. Develop a new generation of artificial intelligence compilers suitable for model training scenarios to achieve automatic operator generation and optimization. Promote the widespread adaptation of artificial intelligence training reasoning chips and framework models, develop an artificial intelligence chip evaluation system, and achieve basic hardware and software automation evaluation.

(11) Explore new paths for general artificial intelligence

Develop a basic theoretical system for general artificial intelligence, strengthen basic theoretical research on artificial intelligence mathematical mechanisms, autonomous collaboration and decision-making, and explore new paths for general artificial intelligence such as general intelligent bodies, embodied intelligence, and brain-like intelligence. Support the research of value- and causality-driven general intelligent entities, build a unified theoretical framework system, rating standards and testing platforms, develop operating systems and programming languages, and promote the application of underlying technology architecture of general intelligent entities. Promote research and application of embodied intelligent systems, break through the perception, cognition, and decision-making technologies of robots under complex conditions such as open environments, generalization scenarios, and continuous tasks. Support the exploration of brain-like intelligence, research on core technologies such as the connection mode, coding mechanism, and information processing of brain neurons, and inspire new artificial neural network modeling and training methods.

IV. Promote the innovation and application of general artificial intelligence technology scenarios

(12) Promote demonstration applications in the field of government services

Around government consultation, policy services, receiving and handling complaints, and government affairs, using artificial intelligence’s advantages in semantic understanding, autonomous learning, and intelligent reasoning, improve the intelligence of government consultation system’s question-answering abilities, enhance the “Jingce” platform’s standardized management and precision service capabilities, assist citizen service hotlines in efficiently responding to citizen demands, and promote accurate guidance and efficient approval of government affairs.

(13) Explore demonstration applications in the medical field

Support research-oriented medical institutions to extract intelligent guidance, auxiliary diagnosis, intelligent treatment and other scenario requirements, fully tap into multimodal medical data such as medical literature, medical knowledge maps, medical images, biological indicators, and collaborate with artificial intelligence innovation entities to develop intelligent applications, achieve accurate identification and prediction of symptoms, signs, and specialized diseases, and enhance the intelligence level of disease diagnosis, treatment, prevention, and full-course management.

(14) Explore demonstration applications in the field of scientific research

Develop scientific intelligence, accelerate artificial intelligence technology’s empowerment of scientific research in new materials and innovative drugs. Support energy, materials, and biology-related laboratories to establish special research cooperation projects, carry out joint R&D with artificial intelligence innovation entities, fully tap into experimental data in the materials, proteins, and molecular drugs fields, develop scientific computing models, carry out prediction of novel alloy materials, protein sequences, and innovative drug chemical structure sequences, and shorten scientific research experiment cycles.

(15) Promote demonstration applications in the financial field

Systematically layout “Ranking First” projects, promote financial institutions to further open industry application scenarios; support financial technology innovation entities to focus on intelligent risk control, intelligent investment consultation, intelligent customer service, etc., develop long text accurate analysis and modeling technology for financial professionals, complex decision-making logic and model information processing fusion technology, support investment-assisted decision-making in the financial field.

(16) Explore demonstration applications in the field of autonomous driving

Support autonomous driving innovation entities to develop multimodal fusion perception technology, based on vehicle-road cooperative data and vehicle driving multi-sensor fusion data, improve the multi-dimensional perception and prediction performance of autonomous driving models, effectively solve complex scenario tail problems, and assist in improving the generalization ability of vehicle-mounted autonomous driving models. Support the open-source of vehicle-road cooperative autonomous driving dataset in the construction of the 3.0 project in the Beijing high-level autonomous driving demonstration zone. Conduct cloud-controlled autonomous driving model testing based on low-latency communication, and explore new technological paths for autonomous driving.

(17) Promote the demonstration and application in the field of urban governance

Support AI innovation entities to combine the needs of smart city construction scenarios, take the lead in applying large-scale model technology in urban brain construction, accelerate the research and development of multi-dimensional perception system fusion processing technology, realize the unified perception, correlation analysis, and situation prediction of the underlying business of smart cities, and provide more comprehensive support for urban governance decisions.

5. Explore and create an inclusive and prudent regulatory environment

(18) Continue to promote regulatory policy and regulatory process innovation

Explore and create a stable and inclusive regulatory environment, encourage innovation entities to use safe and reliable software, tools, computing, and data resources, carry out independent innovation, promotion, and international cooperation of basic technologies such as AI algorithms and frameworks. Strive to take the lead in the core area of Zhongguancun National Independent Innovation Demonstration Zone, and promote the implementation of inclusive and prudent regulatory pilot projects.

(19) Establish a normalized service and guidance mechanism

For AI-related internet information services with public opinion attributes or social mobilization capabilities, carry out normalized contact services and guide innovation entities to introduce technology tools to conduct security testing, declare security assessments in accordance with regulations, and perform algorithm filing procedures.

(20) Strengthen network service security and personal data protection

Guide innovation entities to strengthen network and data security management, implement the main responsibility for network security, data security, and personal information protection, strengthen the construction of security management systems, and work implementation. Encourage innovation entities to carry out data security management certification and personal information protection certification, implement data cross-border transmission security management systems, and comprehensively enhance network security and data security protection capabilities.

(21) Continuously strengthen scientific and technological ethical governance

Strengthen the research of AI ethics and safety norms and social governance practices. Build a public service platform for scientific and technological ethical governance in the field of general artificial intelligence, serve government supervision, and promote industry self-discipline. Carry out scientific and technological ethical review and related business training, and strengthen the awareness of scientific and technological ethical norms of various responsible subjects. Carry out in-depth scientific and technological ethical education and publicity, and build a good atmosphere of scientific and technological ethical governance for artificial intelligence.

We will continue to update Blocking; if you have any questions or suggestions, please contact us!