Generative Artificial Intelligence, particularly Generative Adversarial Networks, has garnered significant attention across diverse sectors, showcasing its ability to create realistic content based on previous examples. While mainstream outcomes of Generative Adversarial Networks rely on human verification —a few new instances are generated to be watched or listened to by humans for checking their realism— the scientific community encounters distinct challenges, especially in Physics. Due to the volume of new instances generated, human-based verification methods may be impractical. This work explores the application of Generative Artificial Intelligence, focusing on conditional Generative Adversarial Networks, to address issues inherent in Astrophysics scenarios where observational data volume is limited, and standalone simulations could be infeasible due to stochastic processes or when examples of interest represent only a fraction of the entire simulation. Human-based supervision of thousands to billions of scientific examples for similarity verification becomes impractical, highlighting a unique barrier of Generative Neural Network in scientific data generation. Generative Adversarial Networks, comprising generator and discriminator networks engaged in a training game, hold promise for generating new data. The discriminator evaluates realism, while the generator aims to produce instances indistinguishable from real data. An extension of this, conditional Generative Adversarial Networks, incorporates additional input (labels) for targeted generation. One the most critical issue of this kind of network is collapse mode, which adversely impacts the quality and diversity of generated instances. In the context of Astrophysics, this work proposes a novel approach. Gaussian Processes are used to evaluate the fit between generated and observed instances, generating similarity measurements and identifying collapse model. This proposal is particularly applied to the generation of galaxies from the Cosmic Evolution Survey (COSMOS).