Yesterday Anne Eisenberg’s story on Robot-Human Cross-Training released in the New York Times. The story does a good job of describing some trends in more advanced human-robot relations such as Baxter – a more flexible and low-cost industrial machine – and future, Internet-capable robots according to Stefan Schaal, who has worked extensively on robots that can mimic human behavior. Yet the most telling aspect of the story comes from a close reading of Prof. Julie Shah’s source research paper, where the experiments are described in detail. What is interesting is just where the complexity of training and human-robot teaming improvement comes from. The robot and human are attempting to collaborate to place three screws in three holes. The human puts the screws in the holes, and the robot screws them in. That’s it. The complexity arises out of the fact that robot and human are so incredibly unfamiliar with the opposite species. The human puts one screw in. “Now, do I wait for the robot to screw it in? Or should I put the second screw in? Wait, now it’s started to screw it down. If I place the second screw, is there a chance it’ll try to screw my hand down? Should I wait for it to finish the first hole? Maybe I should do the third hole first.” This is a study in alien communication: next time you are stranded on an asteroid, and have for company a similarly stuck alien, this is what it will be like trying to bolt your, um, engine back in place so you can take off and come home. No common language, no effective gesturing, no ability to predict how the three-armed bright green alien will behave. In Robot Futures I explain how robots and humans will not be quite so alien to one-another– when robots will be sometimes teleoperated, sometimes autonomous, and we will not even be able to tell which is which. Until then, be patient with clueless robots and clueless aliens.