Many methods based on AE in many areas of science and engineering require high degree of computational power, hence not practically used. The computational requirements of these methods tend to be proportional to the computational requirements of evaluating a single individual. This creates a major bottleneck and also an opportunity for parallelization of the whole process.
With the development of a grid-based AE framework, these methods could be made feasibly.
Having practically feasible AE-based methods, it would be possible, for example, to create more realistic animations, to design better-trained robots, to model any kind of AE systems.