A newly developed artificial intelligence system can now parse raw genetic sequences to reconstruct the evolutionary relationships between species — a capability that would have seemed like science fiction to the biologists who first began mapping phylogenetic trees by hand in the mid-twentieth century.
The achievement builds on a long tradition of computational biology that stretches back to the 1960s, when researchers first began using early computers to compare protein sequences and infer ancestral lineages. What once required years of painstaking manual analysis can now, apparently, be accomplished in a fraction of the time by a model trained to recognize deep patterns within the genetic code itself.
The development fits neatly into a broader arc in AI history: the field's gradual expansion from narrow, rule-based systems into tools capable of engaging with the messy complexity of biological data. Early bioinformatics software could align gene sequences, but lacked the capacity to reason about evolutionary distance at scale. Modern large-scale models, trained on vast genomic datasets, are beginning to close that gap.
Researchers and historians of science will note a certain symmetry here. Charles Darwin sketched his first 'tree of life' diagram in a notebook in 1837 — a rough, intuitive map of species relationships. Nearly two centuries later, machines are drawing those same trees automatically, guided not by intuition but by statistical patterns embedded across billions of base pairs.
The practical implications extend beyond academic curiosity. More accurate evolutionary mapping could accelerate drug discovery, improve our understanding of disease transmission, and shed light on the origins of genetic traits that define life on Earth. As AI tools grow more fluent in the language of DNA, the boundary between computation and biology continues to blur.