The lack of diversity in human genomics research and reference datasets is a known problem, with real-life consequences, but the path forward remains elusive. Solutions to the problem of sampling bias in human genetics research must begin by understanding the following: 1) historical influences and current problems with classification, study design, and conceptual frameworks in human genetics research; 2) tensions between characterizing who has been included in (and excluded from) research, and biases that are introduced through categorizing humans by social and cultural population descriptors (e.g., race, ethnicity, ancestry); and 3) how a lack of diversity and conceptual clarity in human genomics leads to disparities in clinical genetics and precision medicine. Finally, we must recognize how a lack of diversity in the genomics workforce perpetuates these blind spots and how active recruitment and promotion of people from all backgrounds improves both our science and workplace environments.