Recent research used a deep learning model to identify the most important antecedents of unethical behavior from hundreds of potential antecedents measured in the World Values Survey. In the current research, we used a deep learning model to identify the most important outcomes of people’s perceived control. The model identified zero-sum beliefs as one of the most important outcomes (e.g., the idea that people can only get rich at the expense of others; to combatting unemployment, we have to accept environmental problems). Despite extensive research on perceived control and on zero-sum beliefs, no empirical research to our knowledge had explicitly connecting the two. Correlational studies verified that people with low perceived control are more likely to hold zero-sum beliefs. An experiment found that participants who recalled a time in which they had low control endorsed zero-sum beliefs more than participants who recalled a time in which they had high control. These findings suggest that machine learning methods can be used to identify the most important causes of a given outcome and the most important outcomes of a given cause, which can help researchers more quickly and efficiently identify important cause-effect relationships.