This study examines how human capital complements machine learning to mitigate biases. Through experiments with MBA students simulating patent examiner tasks, the researchers find that domain expertise is needed to address biases from incomplete ML inputs, while vintage-specific skills ensure proper use of ML technologies. The hypothesis that ML requires complementary domain expertise and vintage skills to mitigate biases is supported. Managers should consider both attributes of human capital - domain expertise corrects strategic input issues, while vintage skills allow technology operation. Limitations include the student sample and inability to assess long-term impacts.