AI RESEARCH
FluidWorld: Reaction-Diffusion Dynamics as a Predictive Substrate for World Models
arXiv CS.LG
•
ArXi:2603.21315v1 Announce Type: new World models learn to predict future states of an environment, enabling planning and mental simulation. Current approaches default to Transformer-based predictors operating in learned latent spaces. This comes at a cost: O(N^2) computation and no explicit spatial inductive bias. This paper asks a foundational question: is self-attention necessary for predictive world modeling, or can alternative computational substrates achieve comparable or superior results? I.