The introduction of a Cellular Automata (CA)-like structure on the population of Evolutionary Algorithms (EAs) has been verified to be a method to improve solutions quality. However, the study of CA-like structures for Genetic Programming (GP) has been, so far, limited. In this work, we focus on the effect of introducing these structures on Geometric Semantic variants of GP, focusing on the well-known Geometric Semantic GP (GSGP) and its recently introduced variant SLIM-GSGP, which emphasizes producing smaller and more interpretable individuals. Here we provide guidance on how CA-like structures can impact the quality and size of the solutions for GSGP and SLIM-GSGP, giving a clear understanding of the trade-offs involved in applying these methods.