The Japan Science Museum last week revealed a human-like robot called “Alter.”

A neural network controls movements of the robot without needing human assistance. Its movement depends on how many people are around it and change in the room temperature. The robot used its various sensors to detect its proximity, humidity, noise and temperature of the room it is in. Moreover, the robot can sing. The neural network it carries is a learning computer system modeled after the structure of neurons in the human brain. Robotics researchers at labs in Tokyo and Osaka created the robot, which would be on display at the museum through August 6. After the display, the robot will be brought back to the lab for more modification, in other words, to make the robot do even more things.  “This time, Alter doesn’t look like a human. It doesn’t really move like human. However, it certainly has a presence,” said Kouhei Ogawa, one of the researchers behind Alter. When seen in video, the robot gives impression that it is neither a human nor a robot, but is something that is alive. The robot is not first to have a neural network, there are many researchers all over the world such as those working for major companies like Microsoft and Google, also using the same technique of neural nets to make robots for the future. A neural net helps to function and give robot a loose set of parameters to carry out those functions. Researchers could mimic chains of neuron connections by creating artificial neurons. But such nets are not yet perfected, some created have failed and some nets are in progress. According to a report in QZ by Mike Murphy, "Late last week, the museum unveiled its latest installation, a human-like robot called “Alter.” The robot’s movements are entirely governed by a neural network—a learning computer system modeled after the structure of neurons in the human brain—and devoid of human involvement, according to Engadget." Alter has a series of sensors that detect proximity, humidity, noise, and temperature of the room it’s in. Changes in the room’s temperature, or how many people are near the bot, will influence its movements. Kouhei Ogawa, one of the researchers behind Alter, told Engadget: “This time, Alter doesn’t look like a human. It doesn’t really move like human. However, it certainly has a presence.” It’s true. In videos, Alter gives the impression of having something alive in there that’s neither human nor robot. A report published in Popular Mechanics revealed, "The Miraikan, Japan's National Museum of Emerging Science and Innovation, has no shortage of wonders. But its latest temporary exhibit is of particular note: a robot with a neural network, a complex form of artificial intelligence that allows the robot to teach itself and act on its own teaching. It sings, it walks. It is named Alter." This isn't self-awareness, just to be clear. Researchers around the world, including at major companies like Microsoft and Google, are betting on neural nets informing the robots of the future. With a neural net, a robot is given a function and a loose set of parameters with which to accomplish them. By creating artificial neurons, researchers are able to mimic chains of neuron connections. It isn't perfect yet, and many neural net AIs currently fail in their assigned tasks as often as they succeed. But what's important is that, hopefully, they're learning from that failure.