AI trends in 2023: Graph Neural Networks

While AI systems like ChatGPT or Diffusion models for Generative AI have been in the limelight in the past months, Graph Neural Networks (GNN) have been rapidly advancing. In the last couple of years Graph Neural Networks have quietly become the dark horse behind a wealth of exciting new achievements that have made it all the way from purely academic research breakthroughs to actual solutions actively deployed at large-scale. Companies like Uber, Google, Alibaba, Pinterest, Twitter and many others have been already shifting to GNN-based approaches in some of their core products, motivated by the substantial performance improvements exhibited by these methods compared to the previous state-of-the-art AI architectures. Despite the diversity in the type of problems and the differences of their underlying datasets, all these breakthroughs use the single unifying framework of GNNs to operate at their core. This suggests a potential shift of perspective: graph-structured data provides a general and flexible framework for describing and analyzing any possible set of entities and their mutual interactions.