Biologists Question Whether AI Can Decode the Human Genome's Regulatory Complexity

From 1990 to 2003, the Human Genome Project successfully mapped nearly all of the 3 billion chemical components of the human genome, revealing that only around 2 percent are protein-coding genes. A significant challenge lies in deciphering the regulatory functions of the remaining DNA, and some scientists express skepticism about whether AI systems like Evo 2, Genos, and Google DeepMind's AlphaGenome can tackle this issue. Karen Adelman from Harvard Medical School suggests that genes might not be the most fascinating aspect of the genome. Meanwhile, Wendy Bickmore from the University of Edinburgh points out the gaps in knowledge regarding enhancers and cell type data. The piece references Adrian Woolfson's upcoming 2026 book, On the Future of Species, stressing the value of extra-genetic information, while Bickmore remains optimistic about AI's potential to enhance human understanding in this area.