Researchers reported in Science that AI helped create an engineered bacterium partially missing a universally used amino acid building block. The team used AI to propose alternatives to isoleucine in dozens of proteins that comprise bacterial ribosomes. The work represents a test of whether life-like systems can operate with fewer canonical amino acids by leveraging predictive protein structure modeling. While immediate therapeutic applications were not established, the approach is positioned as a pathway toward bespoke protein design and new synthetic biology capabilities. The study adds to the growing use of AI in protein engineering, particularly where changes must remain compatible with core cellular machinery.