We have constructed a generative artificial intelligence model predicting the gravity solutions when the holographic entanglement entropy of the dual quantum field theory is provided as input. The model we utilize is based on the transformer algorithm commonly used in natural language tasks such as text generation, summarization, and translation. The transformer model can understand the implicit relation between input and output sequences by training data. For training, we generate many data sets composed of holographic entanglement entropy and corresponding metric solutions. After training these data, the transformer model predicts the dual geometry from arbitrary test sets of entanglement entropy data. The reconstruction of the dual gravity allows us to get more information on the thermodynamic quantities of thermal systems, which cannot be read directly from entanglement entropy data. In this work, we construct the dual geometry by applying the transformer model. After that, we derive thermodynamic quantities, like temperature and densities of thermal systems, from the entanglement entropy.
Park et al. (Fri,) studied this question.