When people think of self-replicating objects, images of fist-sized robot cubes or sci-fi swarms of nanobots usually spring to mind, but MIT's Self-Assembly Lab is pushing its boundaries with an experiment called Self-Replicating Spheres which eschews mechanical parts entirely.
As with other Self-Assembly Lab projects, including the chair that builds itself we debuted earlier this year, Self-Replicating Spheres is built on customized magnets. When director Skylar Tibbits and his collaborator, Dimitrios Mairopoulos, place the spheres on a table that supplies passive energy, the spheres stick together to form a cell wall-like grouping that grows as researchers "feed" it more spheres. When it reaches critical mass, the "cell" divides into two smaller "cells," which can then replicate again and again as they get more "food."
The effect looks just like those old videos about cell mitosis from middle school science class, but Tibbits makes it clear that this isn't quite biology. "It wasn't necessarily the goal of this to try to create life, it was more like a big questions we all posed to ourselves," he tells The Creators Project. "Is a really elegant pure replication, or mitosis, possible in passive systems with no robotics? Can inanimate objects, like cells, replicate?"
The short answer is, not exactly. The long answer requires us to ask one of the big questions: what is life? "There's not a clear definition of what anyone describes as 'life,'" Tibbits explains. "There are many different definitions, and they all include assembly, replication, metabolism, evolution, growth, all these different characteristics. And what's quite interesting is if you take any of those, like assembly, replication, or growth, we've been able to demonstrate that synthetically outside of biological organisms." So the question is, would putting all these qualities together mean that we succeeded at artificially creating life from scratch?
"The answer is still no," Tibbits says. "We probably wouldn't consider it life, even if everything was the same, but it's all synthetic." Darn. So what's the point of reproducing some parts of of life with man-made stuff, as in Self-Replicating Spheres? He answers, "We're more interested in the principles and the phenomena, and how to reproduce these at larger scale, how to push the boundaries of what's possible, create new properties, new phenomena, and use them in domains we traditionally haven't seen them. In some ways, they might be very similar to life-like or biological systems, but we don't try to argue that it is based on that." In other words, the Self-Assembly Lab thinks of solutions to problems that don't even exist yet, with the natural world as their guide.
Right now, self-replication is right at the top of that list. The concept of tiny robots designed to replicate like living cells dates back to Karel Čapek's 1920 play R.U.R. (Rossum's Universal Robots), but the Self-Assembly Lab draws more inspiration from experiments in the 1950s using wooden blocks designed by psychiatrist, mathematician, and geneticist Lionel Penrose. "We've been thinking about self-replication for a long time," Tibbit says."There's been quite a lot of progress in robotic replication, but it never seemed as elegant as the first time I had seen it."
With less elegant visions of replicating machines in mind, some aren't sure they feel safe with this area of exploration. "The question that a lot of people have is, 'What if it goes out of control? What if you have self-replicating things that take over the world?'" Tibbits says. Combined with AI research at Google, University of Auckland, Harvard, and more, the possibility of a Terminator T-1000 on the loose doesn't seem so far fetched. "We're pretty far from that," Tibbits answers calmly. Whew. "In this case, for example, you have control because you give it food. We're specifically adding food at every step in order to help it keep growing."
Now, the Self-Assembly Lab is working on scaling Self-Replicating Spheres into the hundreds, or even thousands, to see if their magnetic orbs will turn into more complex beings, or just reproduce like rabbits once they've got enough food. But in the long term, Tibbits is already thinking of possible applications. "If you have a system that can control its own growth or performance, or is able to divide itself into two separate parts after a certain amount of time, or understand where an equalibrium is, that's quite an interesting priciple to embed into the everyday world."
Keep up with the Self-Assembly Lab's work on its website.