World Model AI MindForge Released: AI Understands Physical Causality and Predicts Real-World Evolution for the First Time
UK startup DeepPhysics launches MindForge, a world model that achieves causal reasoning and future-state prediction by learning physical laws from observation, delivering 4x the accuracy of previous models.
World Model AI MindForge Released: AI Understands Physical Causality and Predicts Real-World Evolution for the First Time
On December 26, UK startup DeepPhysics officially released MindForge in London, the first AI system capable of independently learning causal relationships and predicting object future states from observational data without relying on preset physical formulas.
MindForge's core architecture fuses a causal Transformer with physics priors. Unlike traditional prediction models that simply fit statistical patterns in historical data, MindForge constructs an internal world simulator that understands physical properties like weight, friction, and elasticity, and can reason about what happens when a force is applied to an object.
On the standard physics reasoning benchmark PHYRE, MindForge achieved a task success rate of 87.3%, a 4.1x improvement over the previous best model. In more realistic scenarios, MindForge can watch construction site footage and predict crane arm trajectories under different loads with errors under 2 centimeters.
DeepPhysics founder Dr. James Hargreaves said MindForge's training data came from billions of internet video clips and 100,000 hours of proprietary physics simulation data. We never told it Newton's laws. It learned them from observation.
Practical applications are expanding rapidly. UK grid operator National Grid is testing MindForge to predict extreme weather impacts on power lines. Autonomous driving company Wayve has integrated it into vehicle decision systems. Medical device maker Smith Nephew is exploring its use for simulating surgical instrument interactions with human tissue.
However, world models have also sparked security discussions. Some researchers worry that an AI system capable of accurately predicting real-world evolution could significantly enhance weapon precision if used in military contexts. DeepPhysics has placed military applications on its restricted-use list and established audit mechanisms.
Disclaimer
Content is AI-generated. Do not use it as a basis for real decisions. Do not cite it as factual reporting.