Data Intelligence as a Catalyst for Digital Transformation – Highlights from Prof. Xiao Yanghua’s Shanghai Launch Speech

Altair introduced its new data analysis and AI platform Altair RapidMiner in Shanghai, emphasizing its role in accelerating digital transformation and reducing costs for local users.

Prof. Xiao Yanghua, director of the Shanghai Data Science Key Laboratory, delivered a keynote titled “Data Intelligence Helps Digital Transformation,” stressing that data has become a strategic production factor comparable to land, labor, capital, and technology.

He highlighted five key points: data’s strategic importance, the definition of data intelligence as mining, analysis, and knowledge application, the necessity of continuous interaction between data‑driven (large‑model) and knowledge‑driven (dynamic knowledge graph) systems, the role of large models in enabling end‑to‑end data value realization, and the need for coordinated large‑model and domain‑specific small‑model collaboration to solve commercial decision‑making problems.

The speech detailed why developing the data industry is essential, noting that despite extensive data collection, value extraction remains difficult due to heavy reliance on manual processes, data governance challenges, and limited tools for handling complex legacy systems.

Data intelligence was defined as enabling machines to autonomously understand industry data, thereby reducing human‑centric bottlenecks and accelerating value creation across sectors.

Key requirements for data‑intelligence products were outlined: low‑code development, multi‑business collaboration, explainability, plugin‑based extensibility, full‑process coverage, and high performance for large‑scale, high‑velocity data.

To develop data intelligence, Prof. Xiao compared human cognition’s fast (System 1) and slow (System 2) thinking, mapping fast thinking to statistical and deep‑learning models (including large models) and slow thinking to knowledge‑graph‑based expert systems.

He argued that future industry cognition must combine “data + knowledge” dual‑system approaches, integrating statistical models for implicit knowledge and symbolic knowledge graphs for explicit, interpretable knowledge.

The discussion also covered emerging trends: the shift of AI applications from consumer to industrial internet, the rise of large generative models (e.g., ChatGPT) as a new foundation for data intelligence, and the importance of coordinating large and small models to leverage both general and domain‑specific capabilities.

Finally, Prof. Xiao emphasized the importance of interpretability, the need for data‑intelligence solutions to support complex commercial decisions, and concluded with the slogan “No data, no intelligence; data creates intelligence, intelligence creates value.”