Total disk arthroplasty (TDA) using an artificial disk (AD) is an attractive surgical technique for the treatment of spinal disorders, since it can maintain or restore spinal motion (unlike interbody fusion). However, adverse surgical outcomes of contemporary lumbar TDAs have been reported. We previously proposed a new mobile-bearing AD design concept featuring a biconcave ultrahigh-molecular-weight polyethylene (UHMWPE) mobile core. The objective of this study was to develop an artificial neural network (NN) based multiobjective optimization framework to refine the biconcave-core AD design considering multiple TDA performance metrics, simultaneously. We hypothesized that there is a tradeoff relationship between the performance metrics in terms of range of motion (ROM), facet joint force (FJF), and polyethylene contact pressure (PCP). By searching the resulting three-dimensional (3D) Pareto frontier after multiobjective optimization, it was found that there was a “best-tradeoff” AD design, which could balance all the three metrics, without excessively sacrificing each metric. However, for each single-objective optimum AD design, only one metric was optimal, and distinct sacrifices were observed in the other two metrics. For a commercially available biconvex-core AD design, the metrics were even worse than the poorest outcomes of the single-objective optimum AD designs. Therefore, multiobjective design optimization could be useful for achieving native lumbar segment biomechanics and minimal PCPs, as well as for improving the existing lumbar motion-preserving surgical treatments.