Drug discovery and development are one of the most challenging, costly, and time-consuming endeavours in modern science. 12–15 years and more than $2.5 billion may be required to progress a single lead compound from bench to clinic through traditional approaches. A very different paradigm, generative artificial intelligence (AI) is revolutionizing this process through accelerated navigation of chemical space, improved lead optimization speed, and new molecular design. Generative AI techniques, such as transformer-based models, generative adversarial networks (GANs), variational autoencoders (VAEs), and recurrent neural networks (RNNs), have previously unheard-of potential for de novo drug discovery, producing new chemical entities with desired therapeutic qualities. These approaches are complemented by knowledge-enhanced frameworks that integrate structured biological information to guide molecular generation. Alongside further maturation, data quality, interpretability of the model, and experimental evidence issues need to be resolved to realize generative AI's potential for revolutionary drug development and, through quicker release of better, more efficacious therapy, improve patient outcomes.