Hughes’s student Maridel Fredericksen used a special microscope to julienne infected ants into slices that were just 50 nanometers thick—a thousandth of the width of a human hair. She scanned each slice, compiled the images into a three-dimensional model, and painstakingly annotated which bits were ant and which bits were fungus. It took three months to mark up just one muscle. To speed things up, Hughes teamed up with computer scientist Danny Chen, who trained an artificial intelligence to distinguish ant from fungus.