Why AI Isn’t Ready to Tackle Our Convoluted Genomes

The article argues that artificial intelligence may help predict how DNA affects biological processes, but it is not yet able to explain how the human genome works. Gene regulation is far more complicated than a simple code: cells switch genes on and off in different combinations, depending on their type and changing surroundings.

The Human Genome Project mapped nearly all of the roughly 3 billion DNA building blocks, but only about 2 per cent of the genome consists of protein-coding genes. Humans have around 20,000 genes, and researchers increasingly focus on how noncoding DNA helps regulate them. Models such as Evo 2, Genos and AlphaGenome learn patterns from genomic data, but their predictions may not provide the deeper understanding of the overlapping controls that govern gene activity.