Matthew Hutson in The New Yorker:
Each year, researchers from around the world gather at Neural Information Processing Systems, an artificial-intelligence conference, to discuss automated translation software, self-driving cars, and abstract mathematical questions. It was odd, therefore, when Michael Levin, a developmental biologist at Tufts University, gave a presentation at the 2018 conference, which was held in Montreal. Fifty-one, with light-green eyes and a dark beard that lend him a mischievous air, Levin studies how bodies grow, heal, and, in some cases, regenerate. He waited onstage while one of Facebook’s A.I. researchers introduced him, to a packed exhibition hall, as a specialist in “computation in the medium of living systems.”
Levin began his talk, and a drawing of a worm appeared on the screen behind him. Some of the most important discoveries of his career hinge on the planarian—a type of flatworm about two centimetres long that, under a microscope, resembles a cartoon of a cross-eyed phallus. Levin is interested in the planarian because, if you cut off its head, it grows a new one; simultaneously, its severed head grows a new tail. Researchers have discovered that no matter how many pieces you cut a planarian into—the record is two hundred and seventy-nine—you will get as many new worms. Somehow, each part knows what’s missing and builds it anew. What Levin showed his audience was something even more striking: a video of a two-headed planarian. He had cut off the worm’s tail, then persuaded the organism to grow a second head in its place. No matter how many times the extra head was cut off, it grew back.
The most astonishing part was that Levin hadn’t touched the planarian’s genome. Instead, he’d changed the electrical signals among the worm’s cells. Levin explained that, by altering this electric patterning, he’d revised the organism’s “memory” of what it was supposed to look like. In essence, he’d reprogrammed the worm’s body—and, if he wanted to, he could switch it back.