New Apple study challenges whether AI models truly “reason” through problems

AI researcher Gary Marcus, who has long argued that neural networks struggle with out-of-distribution generalization