Deloitte China Board Chair Jiang Ying recommends: Promoting the development of high-quality datasets in three areas

China Economic Net, Beijing, March 8th – (Reporter Li Yuehua, Ma Changyan) “Currently, China has made significant progress in the field of artificial intelligence, but to shift from ‘catching up’ to ‘leading’, we must rely on high-quality datasets to support algorithm development and application implementation,” said Jiang Ying, member of the National Committee of the Chinese People’s Political Consultative Conference and chairman of Deloitte China, in an interview with China Economic Net reporters.

High-quality datasets are the key foundation for driving the large-scale application of artificial intelligence and are of great significance for enhancing the country’s technological innovation capabilities and promoting high-quality economic development. In recent years, the scale advantage of our country’s data resources has continued to expand, and the activity level of data resource development and utilization has steadily increased. According to data, by 2025, our country will have built over 100,000 high-quality datasets, with a scale exceeding 890PB (petabytes).

Jiang Ying pointed out that high-quality datasets can be directly used for the development and training of artificial intelligence models, effectively improving model performance. However, she found in her research that there are still issues in the current construction of datasets in our country, such as inconsistent standards, weak management services, and insufficient application orientation, leading to redundant construction of datasets, low reuse rates, and limited value release.

In response, Jiang Ying proposed three suggestions: first, to establish a unified standard system covering classification, metadata, and quality assessment to promote the standardized transformation of existing data; second, to establish a special coordination mechanism that provides standardized processes, tool templates, and technical support, as well as to lower the burden on enterprises through the establishment of financial incentives; third, to strengthen application orientation, requiring project approvals to clearly define usage scenarios, and to ensure that construction results match demand through review and supervision, while promoting mature datasets to be included in public platforms for sharing and reuse.

Regarding the application-oriented construction mechanism for high-quality datasets, Jiang Ying further stated that during the construction process, there should be periodic reviews, acceptances, and dynamic supervision to continuously calibrate the direction of dataset construction, ensuring that construction results are highly aligned with actual application needs, and avoiding “building for the sake of building.”