Modeling-to-generate alternatives (MGA) is a technique for using mathematical programming models to generate a small number of different solutions for the decision maker to consider when dealing with complex, incompletely defined problems. The logic of MGA is presented in the context of concerns about the limitations of mathematical models and the human decision-makers who use them. Arguments and experimental evidence are presented to support the assumption that the human-machine decision-making system will perform better when the human is presented with a few different alternatives than when presented with a homogeneous set of alternatives, as might result from sensitivity analysis.