Translated data: After 35 years of debate, researchers from NYU and Pompeu Fabra University have first demonstrated that neural networks possess systematic generalization capabilities akin to humans. Their proposed MLC model, through training in the task environment, enables neural networks to gradually acquire systematic reasoning abilities in understanding the meanings of combined words, surpassing GPT-4. This research will profoundly impact the fields of artificial intelligence and cognitive science, endowing neural networks with systematic generalization capabilities.