A new article in *Google Research’s genomics division and a feature in *New Atlas note that the ambitious Earth BioGenome Project (EBP)—which aims to sequence the genomes of roughly 1.85 million eukaryotic species—is getting a boost from artificial intelligence tools developed by Google. These AI systems help accelerate and improve the accuracy of genome assembly, variant calling and polishing, reducing both the time and cost of sequencing.
Google’s tools such as DeepVariant (released 2018) and DeepPolisher are cited as key enablers. DeepVariant treats sequencing-read alignment as an image-classification problem, allowing neural networks to distinguish true genetic variants from sequencing errors. Meanwhile, DeepPolisher targets genome assemblies—reducing error rates by roughly 50 % in some tests, enabling more complete and accurate reference genomes for plants, animals and fungi.
Despite the technological advances, the scale of the task remains enormous. Although EBP has made progress, it still has sequenced only a small fraction of the nearly 2 million target species. The article underscores that AI and automation are making the leap from human-centred manual pipelines toward high-throughput, end-to-end systems that may make such a “moonshot for biodiversity” feasible. But it also flags that sequencing, assembling and annotating species deeply requires field sampling, biobanking, global collaboration, and sustained funding.
In short, AI is not just accelerating human genomics—it’s being marshalled to bring the full tree of life into the digital realm. The implications span conservation (helping identify genetic diversity of endangered species), biotechnology (unlocking new genetic resources), and fundamental biology (understanding evolutionary history). But as the article warns, the promise comes with logistical, ethical and ecological challenges that must be addressed alongside the technology.
