In this paper the procedures for determining the anisotropy parameters of Hill-48 and Karafilis-Boyce stress functions are considered. By conventional calibration procedures, for Hill-48 stress function the parameters are determined based on simple analytical expressions, while for Karafillis-Boyce stress function they are determined by some iterative procedure. By this approach, if non-associated flow rule is assumed, seven experimental data are utilized in the model calibration: four yield stresses and three Lankford coefficients. In order to introduce more experimental data in the model calibration, this paper considers the use of genetic algorithms in determining the parameters of Hill-48 and Karafillis-Boyce stress functions for aluminum alloy sheet AA2090-T3. The considered calibration approach, for the selected material, uses eight values of yield stresses and eight values of Lankford coefficients. For the stress functions calibrated by this approach, the predictions of orientational dependences of yield stress and Lankford coefficient as well as the contours for the different shear stress values are considered. For Karafillis-Boyce yield function, the impact of the genetic algorithm settings on the parameter's solutions was additionally considered. When determining the parameters of Hill-48 and Karafillis-Boyce plastic potential, the constraint for the shear stress state is introduced. Considering the introduced shear constraint, the predictions of plastic flow for different stress states were additionally analyzed. Constitutive models that utilize Hill-48 stress function with parameters determined by applying genetic algorithms, without and with the shear constraint, were also analyzed in predicting the cylindrical cup deep drawing process for AA2090-T3 sheet.