Technical University of Munich researchers demonstrated that artificial neural networks trained with biological data—specifically spontaneous retinal waves observed in early visual system development—show faster and more accurate prediction of movements than networks trained without such pre-training. Retinal waves comprise patterned neural activity preceding eye opening, coordinating retinal-brain wiring. Incorporating these signals into AI training improved simulated navigation tasks and movement predictions from real-world footage, suggesting that embedding biological development principles into AI architectures can advance autonomous systems and robotics.