Gartner releases China's data analytics and artificial intelligence technology maturity curve in 2023

Gartner predicts that by 2026, more than 30% of white-collar jobs in China will be redefined, and the skills to use and manage generative AI will be very popular.

Gartner’s 2023 Hype Cycle for Data Analytics and AI in China reveals four fundamental themes related to data, analytics, and AI in China: China data strategy that prioritizes business outcomes, regional data and analytics, and AI Ecosystems, the collapse of data centers, and artificial intelligence becoming the new symbol of national power.

In this curve, the largest number of technologies are about to enter the period of expected expansion. Zhang Tong, senior research director at Gartner, said: “Innovation is often touted as a solution to traditional bottlenecks and is expected to solve common concerns of Chinese CIOs, such as hardware resource shortages, scalability, sustainable operations, security risk mitigation, and technological independence. control and multi-domain applicability of AI models, resulting in clear business value. However, end users value tangible impact more than abstract strategic concepts."

Source : Gartner (August 2023)

Data Weaving

Data weaving is a design framework for obtaining flexible and reusable data pipelines, services, and semantics involving Data integration, active metadata, knowledge graphs, data profiling, machine learning and data classification. Data weaving subverts the existing dominant approach to data management. It is no longer "tailor-made" for data and use cases, but "observation first and then use".

Zhang Tong, senior research director at Gartner, said: "The emergence of data, analytics and AI use cases, as well as rapidly changing data security regulations, have led to the complexity and uncertainty of data management in China. Data weaving can make full use of Sunk costs, while also providing prioritization and cost control guidance for new expenditures on data management infrastructure."

Data Asset Management

Data asset management refers to the management, processing and utilization of The process of generating data that is a valuable asset to business operations. Data asset management applies to a variety of data forms - for example, images, videos, files, materials and transaction data in the system, and covers the entire data life cycle from data acquisition to destruction. The purpose is to manage data in the same way as assets. and create value from it.

Data, as a new production factor, has become a competitive advantage for enterprise organizations. Data is fast, diverse, voluminous and factual, so organizations must integrate processes to generate data insights.

Zhang Tong, senior research director at Gartner, said: "Data assets can not only improve operational quality and decision-making levels, but also create more business value. They can also generate new business models and use data to directly monetize. However, despite Value creation is accelerating, but data assets still have potential risks. Enterprise organizations must manage data assets carefully to avoid regulatory violations and accidental data leakage."

assembled data and analysis

assembled data and Analytics (D&A) leverages container- or business microservices-based architecture and data weaving concepts to assemble existing assets into flexible, modular, and user-friendly data analytics and artificial intelligence (AI) capabilities. This technology can use a series of technologies to transform data management and analysis applications into data analysis and AI components or other application modules with the support of low-code and no-code capabilities, and support adaptive and intelligent decision-making.

Faced with the rapidly changing business environment, Chinese enterprises and institutions need to improve their agility and speed up the output of insights. Assembled D&A helps enterprise organizations use modular data and analysis capabilities to integrate multiple insights and reference information into various measures to avoid fragmented development. Enterprise organizations can further improve delivery flexibility by assembling or reorganizing D&A capabilities to cope with different usage scenarios.

Large model

Large model is a large parameter model trained in a self-supervised manner on a wide range of data sets, most of which are based on Transformer architecture or diffusion deep neural network architecture. And may become multimodal in the near future. The name Big Model comes from its importance and broad suitability for a variety of downstream usage scenarios. This ability to adapt to a variety of scenarios benefits from sufficient and extensive pre-training of the model.

Large models have now become the preferred architecture for natural language processing and have been applied in computer vision, audio and video processing, software engineering, chemistry, finance and law. A popular sub-concept derived from large models is large language models based on text training.

Zhang Tong, senior research director at Gartner, said: "Large models have the potential to provide enhanced effects for applications in various natural language use cases, and therefore will have a profound impact in vertical industries and business functions. They can improve employee productivity. , automate and enhance customer experience, and cost-effectively create new products and services to accelerate digital transformation."

Data Middle Office

Data Middle Office (DMO) is a practice of organizational strategy and technology. Through the data center, users in different business lines can efficiently use enterprise data to make decisions based on a single source of truth. Creating a data center can be a way to build assembleable and reusable data and analysis capabilities for enterprises. These capabilities can provide unique digital operations and integrate digital operations throughout the entire value chain through the technology stack.

The reason why many Chinese companies adopt data middle-end practices is to reduce the technical redundancy of their data and analysis architecture, open up data islands of different systems, and promote reusable data and analysis capabilities. However, the data center has in many cases failed to deliver on its promise of assembled agile D&A capabilities, and thus its position in the market has been weakened. Many organizations and vendors are reluctant to adopt this concept internally, or simply remove it from their promotion.

The above is the detailed content of Gartner releases China's data analytics and artificial intelligence technology maturity curve in 2023. For more information, please follow other related articles on the PHP Chinese website!