According to a recent news release, the researchers gave the system a prompt to design a robot that could walk on flat surfaces, and after running through several iterations, the algorithm produced a functional design in a matter of seconds.
The announcement said the AI program, cited in an Oct. 3 study in the Proceedings of the National Academy of Sciences (PNAS), can run on a lightweight personal computer and design novel structures whole cloth, whereas many other AI systems need supercomputers and larger data sets to function and still can’t generate entirely new ideas.
“We discovered a very fast AI-driven design algorithm that bypasses the traffic jams of evolution, without falling back on the bias of human designers,” said Sam Kriegman, an assistant professor of computer science, chemical and biological engineering, and mechanical engineering, who led the team’s work. “We told the AI that we wanted a robot that could walk across land. Then we simply pressed a button and presto! It generated a blueprint for a robot in the blink of an eye that looks nothing like any animal that has ever walked the earth. I call this process ‘instant evolution.’”
According to the news release, Kriegman gained media attention in 2020 for developing xenobots, the first robots made entirely from biological cells. It said Kriegman’s current research team hopes to use their new AI program to “explore the potential of artificial life.” While the AI-designed robot itself is fairly crude, Kriegman said, he believes their AI program could mark a new era of AI-designed tools.
“When people look at this robot, they might see a useless gadget,” Kriegman said in the announcement. “I see the birth of a brand-new organism.”
[Evolution] cannot see into the future to know if a specific mutation will be beneficial or catastrophic. We found a way to remove this blindfold, thereby compressing billions of years of evolution into an instant.
Sam Kriegman, an assistant professor of computer science, Northwestern University
“Now anyone can watch evolution in action as AI generates better and better robot bodies in real time,” Kriegman said in an announcement. “Evolving robots previously required weeks of trial and error on a supercomputer, and of course before any animals could run, swim or fly around our world, there were billions upon billions of years of trial and error. This is because evolution has no foresight. It cannot see into the future to know if a specific mutation will be beneficial or catastrophic. We found a way to remove this blindfold, thereby compressing billions of years of evolution into an instant.”
While the AI’s first robot can do little more than shuffle forward, Kriegman hopes the team’s research will lead to the development of similar robots that are able to navigate the rubble of a collapsed building, or follow thermal and vibrational signatures to search for trapped people and animals, among other uses. The AI may also be able to design nanorobots for medical applications, he added.
“The only thing standing in our way of these new tools and therapies is that we have no idea how to design them,” Kriegman said. “Lucky for us, AI has ideas of its own.”