Evolutionary algorithms were proposed to automatically find solutions to computational problems, much like evolution discovers new adaptive traits. Lately, they have been used to address challenging questions about the evolution of modularity, the genetic code, communication, division of labor and cooperation. Evolutionary algorithms are increasingly popular in biological studies, because they give precise control over the experimental conditions and allow the study of evolution at unprecedented level of detail. Nevertheless, evolutionary algorithms have their own caveats, which are often overlooked. Here, we highlight one of them by exposing a terminological conflict between definitions of fitness used in biology and in evolutionary algorithms.