Credit: Patterns (2025). DOI: 10.1016/j.patter.2025.101392 Language models are now also being used in the natural sciences. In chemistry, they are employed, for instance, to predict new biologically active compounds. Chemical language models (CLMs) must be extensively trained. However, they do not necessarily acquire knowledge of biochemical relationships during training. Instead, they draw conclusions based on similarities and statistical correlations, as