Researchers in Cambridge and Zurich have programmed a robot to continuously rebuild and create a better robot.
It’s being popularly described as “having babies”, but in reality, it’s much more advanced, because you have to wait for babies to grow up.
This machine builds a structure out of blocks and motors, then makes it better. It even uses glue:
It then turns them on, and a camera detects how far the blocks travel:
Using this data, the mother robot pulls them all back and rearranges the blocks into “fitter” machines.
Fitness was defined as the distance the child bot travelled in eight seconds. That had to be cut to four seconds in the last two experiments, as the mutated “child” had begun moving too quickly.
Here’s a description of the selection process which would send chills down the spines of members of the anti-AI movement:
After the fitness evaluation of one generation is completed, the agents’ genomes for the following generation are created. The elite of fittest individuals advance unchanged to preserve their abilities.
The remaining slots of the next generation are filled with agents whose genomes were altered through mutation or crossover.
Here you can see how the robot analyses the performances of its creations:
Luzius Brodbeck, one of the researchers from the Institute of Robotics and Intelligent Systems at ETH Zurich, told CNBC the robots could be used in remote areas or for disaster response.
But that’s unlikely to reassure any of the 1000-plus leading robotics and AI experts who recently signed a letter to the UN urging a ban on lethal autonomous weapons systems.
Elon Musk in particular has been leading the charge against going down the AI road too far. In January, he donated $US10 million to the Future of Life Institute (FLI) to fund a program with the goal of making sure AI doesn’t completely overrun our ability to regulate it and end up destroying us all.
A robot that can rearrange other robots into better robots might, to Musk, sound like something that needs to be carefully controlled. But Brodbeck isn’t as concerned.
“I think it makes sense to think about this,” he told CNBSC, “but I personally am not afraid that robots will take over the world.”
You can read the full paper at PLOS ONE. Here’s the robot in action: