CityU Hong Kong Researchers Use AI to Confirm Water's 2-State Model
Updated
Updated · BusinessLine · Jun 29
CityU Hong Kong Researchers Use AI to Confirm Water's 2-State Model
2 articles · Updated · BusinessLine · Jun 29
Summary
City University of Hong Kong researchers say AI analysis has found evidence that water constantly switches between 2 microscopic structures—a high-density form and a low-density form.
Millions of data points from simulations of thousands of water molecules were fed into an unsupervised deep-learning system, which identified patterns scientists say support the long-debated two-state model.
The finding aims to explain anomalies such as water reaching maximum density near 4°C, then expanding as it cools further, along with its unusual heat storage and pressure behavior.
Researchers said the AI completed in about 18 months analysis that might otherwise have taken years, pointing to a broader role for machine learning in resolving longstanding scientific puzzles.
If AI can reveal the hidden secrets of water, what practical new technologies will this breakthrough unlock?
After an AI decoded water's two-state mystery, what other scientific puzzles are on the verge of being solved?
Will AI's new understanding of water's dual nature help us create entirely new materials from scratch?
Groundbreaking AI Study Confirms Water’s Two-State Model Using 74 Million Molecular Configurations
Overview
In June 2026, researchers at City University of Hong Kong, led by Professor Xiao Cheng Zeng, made a major breakthrough by using artificial intelligence to confirm the long-debated two-state model of water. This model suggests that liquid water is not uniform but a mix of two different local structures: one denser and disordered, the other less dense and more ordered. For decades, direct molecular evidence for this idea was missing. The team trained an advanced AI on about 74 million water molecule configurations from accurate molecular simulations, finally providing strong molecular-level proof for water’s dual nature.