35 major state-owned enterprises test causal world model in production
Chinese company Zhongshu Ruizhi has launched a causal world model system that helps enterprises in energy, manufacturing, finance and other high-stakes industries make better decisions by reasoning through cause and effect, not just correlation. The system has been deployed at over 35 large state-owned enterprises, covering more than 800 business scenarios, and reduces false alarms while improving risk detection timing.
Zhongshu Ruizhi, founded in 2020, specializes in enterprise-level complex intelligent decision-making, primarily serving large state-owned enterprises. During WAIC, the company released its 'AI for Reasoning' causal intelligence system, aimed at addressing the shortcomings of existing AI in core business decisions. The system is built on a 'causal world model' that incorporates dynamic causal graphs, which can be automatically generated from enterprise documents and data, and continuously evolve. Applied to scenarios like oil and gas drilling well control, the system can provide early warnings 15-20 minutes ahead of dangerous conditions, achieve 94% accuracy in root cause localization, and filter out over 40% of false alarms. The system has been deployed at more than 35 large state-owned enterprises, running for over 15,000 hours without a single model hallucination causing a decision error, covering over 800 core business scenarios. It supports both SaaS subscription and private deployment. Zhongshu Ruizhi emphasizes that every prediction includes a time limit, is logged, and is verified against actual outcomes.
- Сокращения
- WAIC = World Artificial Intelligence Conference — Всемирная конференция по искусственному интеллекту
Source: QbitAI 量子位 —
original