A few days ago, I was watching an episode of The Age Of AI on YouTube Originals. It is a fascinating series for AI developers. In the episode, How Far Is Too Far, an AI technology called BabyX astonished me. BabyX is a research project which aims to create a lifelike “baby” through the means of affective computing.
So, BabyX is a project of Auckland’s Bioengineering Institute Laboratory for Animate Technologies. Basically, they are creating a human-like simulation of a human baby that interacts and learns from its environment.
Just as a human has five senses, BabyX is provided with cameras and microphones so that it can interact with its operator and learn from them. All the data which is coming from its senses or more precisely sensors come into a huge Artificial Neural Network. The institute has designed a neural network inspired by our biological neural cortex. Just as a human baby learns from imitation, BabyX tends to simulate the learned actions and even the minor emotions found on our face.
As seen in the video, a camera just acts as an eye for BabyX. It can detect facial emotions and interpret them with their NN.
BabyX’s operator frightens it by making a loud cry and the Baby begans to cry! See here.
This could be our first step in making machines more friendly as they can understand as well as show human-like emotions.
As discussed earlier, BabyX has a brain of its own, made from Artificial Neurons which resembles the human brain. It receives data from the various sensors and tries
The human brain consists of 85 billion neurons. With years of research, we have tried to mimic these biological neurons to create artificial neurons. A biological neuron
Biological neurons do three things. First, they get a signal through dendrites. Second, this signal is processed by the soma and then passed through the axon and finally, this signal reaches another neuron present in the network with the help of axon terminals.
Out of 85 billion if we are even able to create an NN with a few million neurons, then we can program machines that have more or less human abilities.
In the above scenario, if we are able to build an NN with nearly 85 billion neurons connected via senses ( cameras, microphones, etc. )? What will we see? Most probably a human-like software that is really alive and interacting with its creators.
It will be more like a human toddler but without physical existence, as we have. A Baby whose consciousness is precisely defined by mathematics, who gets alive when electrical signals are provided and dies as soon as the power is gone. A consciousness, a state of being alive and interacting with the environment in a world of 0s and 1s. There’s nothing more fascinating than this?
This is going to change the phase of humanity such as fire and the wheel did.
This was a short story of BabyX as I found quite insightful and interesting. Thanks for reading and Happy Machine Learning!