Recently, the AI Data Industrial Base of Ant Digital officially started operations in Taicang Zhihui Valley · Science and Technology Innovation Park. As the first AI data industrial base in Suzhou, this base aims to provide scenario-based, high-quality training data support for the deployment of large models in fields such as manufacturing, finance, and healthcare in the Yangtze River Delta region through cross-industry talents and cutting-edge technologies.
Data is regarded as the core textbook for AI learning. However, when AI is deployed in complex scenarios such as industry, it often faces bottlenecks such as a lack of industry knowledge and scarce high-quality data, which urgently requires the participation of professional talents in construction. Professional AI data annotators, known as "AI instructors," ensure that training data is accurate and free from bias through precise annotation and correction, thus improving the safety and reliability of model outputs, and continuously "teaching" AI to understand semantics and be more human-like.
Zhao Qiuyan graduated from a university with a bachelor's degree in Financial Studies. She used to work as an accountant and is skilled at correcting AI answers related to financial issues. Yan Fengna, who studied law, mainly works on labeling "red line" issues in the text output by large models, teaching AI "what can be said and what cannot be said." Li Ning, an AI annotator, used to be an art teacher. He needs to deal with a large amount of image data and label "content risks" within them. Professionals from fields such as medicine, foreign languages, and tourism are also joining the data annotation work for various vertical models.
"The average educational background of the AI annotators at Ant Digital is above bachelor's degree. The Taicang government has provided strong support in areas such as park construction, talent recruitment, and the entry of ecological enterprises," introduced Zhang Kai, General Manager of the AI Technology Business Unit of Ant Digital. Ant Digital has already launched industry-academia integration programs with multiple universities in Jiangsu to cultivate practical AI talents. According to the plan, the base will expand to a scale of 300 people in the future.
On the technical level, Ant Digital's self-developed AI data annotation platform can intelligently recommend annotation content and automatically complete tags, combining automated quality inspection with manual sampling to ensure data quality. Currently, its annotation services cover dozens of scenarios such as financial risk control, content review, autonomous driving, and intelligent customer service, providing a reliable data foundation for the industrialization of large models in multiple fields.