When it comes to drug discovery using compounds, there have been “first resources (libraries)” derived from natural products such as microorganisms and plants, and “second resources (libraries)” derived from libraries of compounds conceived and synthesized by humans. Now, attention is focused on “third resources” derived from AI-generated latent compound spaces. While the first and second resources are depleted, the synthesis of potential compounds by AI, opens up a vast new space, offering the promise of new discoveries. Until now, drug discovery has followed the Design-Make-Test-Analyze (DMTA) cycle, but the Generate-Make-Test-Learn (GMTL) cycle will become central in the future