Last week, Kokoro Co. Ltd. released video of the latest iteration of its work in robotics, taking another step in the process of bridging the “uncanny valley”—the idea that the closer that robots and other non-human objects approach to looking authentically human, the stronger a reaction of revulsion, fear, or mistrust they inspire in their human observers. Kokoro is well-known in the robotics field for creating lifelike humanoid robots capable of recognizing and mimicking human facial expressions and body language. You may recognize the Kokoro name in reference to the I-Fairy, a humanoid robot that presided over a wedding in Japan last summer, and the Geminoid project, where customers can have themselves reproduced in silicone and wire.

The Actroid-F (the ‘F’ stands for female) robot was built and programmed to monitor patients in a hospital setting. Currently, programmers are teaching the robot to mimic patients’ facial expressions, and to recognize the differences in their smiles and grimaces. The result is a very lifelike robo-nurse that can be used to monitor the feelings and needs of hospitalized patients. On the surface, this seems like a good idea; we already use machines to monitor patient vital signs and administer life-saving medicines, so why not use machines to monitor patient morale?

The question I want to pose, however, is why this machine has been given human form. Medical professionals don’t generally feel the need to paste googly eyes on IV drips, or put a mustache on the X-ray machine before scanning a patient. So why should a technology meant to monitor patients’ faces have to look human? If monitoring faces was the only goal here, a webcam and some facial expression recognition software would suffice, and would probably do the job in a much less intrusive or disturbing way. Is there something intrinsic to interacting with machines given human form that would somehow improve the performance of the device?

And what about the fact that the robo-nurse in question is structured to look not just human, but female? This seems to have its roots in normative expectations of what gender a nurse should be, but remember: this isn’t a nurse, it’s a machine. Giving it human form is one thing, but why must it be female? And beyond making the machine female, why must the body we give it be an attractive, normative body?

I’m reminded of Kelly Joyce’s Magnetic Appeal, and her discussion of how trends toward visuality have led doctors, technicians, and patients to give the magnetic resonance imaging (MRI) machine so-called non-human agency in the scanning and diagnosis of patients. According to Joyce, modern patients put more stock in a visual representation of their health—usually in the form of a magnetic resonance image—than they do in physician opinion. She asserts that medical experts are not immune to this preference; in using the MRI machine to image and assess patients’ health, physicians and technicians will contribute all of the diagnostic and healing power of the MRI to the machine, often valuing the machine’s output more than their own initial assessment. In essence, the tools build the house despite the carpenter, and not because of him.

Is Actroid-F the next step in visuality and non-human agency? Are we now so comfortable giving machines agency in human processes that it now actually makes sense to give them human form?