A new discipline wants to study the interactions of AI systems in the same way that it analyzes behavior in humans.
Artificial intelligence and machine learning are based on algorithm that determine the responses of the machines.
For humans it is sometimes difficult to understand what reasons lead to algorithm to choose one answer or another.
A group of researchers has decided that scientific methods that study artificial intelligences based primarily on these logarithms are not entirely effective. This group of innovative scientists proposes to create a new academic discipline to be known as Machine behaviour (Machine Behaviour). His proposal is based on studying AI systems in the same way that animals and humans are studied, through empirical observations and experimentationaccording to an MIT article.
The person in charge of studying this field would be a computer engineer what a sociologist is a neuroscientist. The first analyzes how an agent (artificial or biological) behaves in its habitat and interacts with other agents. The second seeks to understand the mechanics behind these behaviors.
Iyad Rahwan, researcher of the report, believes that there is an increase in autonomous and decision-making machines, and therefore they should be investigated as a new actor with their own patterns of behavior.
Do not panic, this does not mean that machines have free will. They simply try to stop seeing machines as passive subjects.
What would be the work of a Machine Behavior scholar?
Their roles could be to interrogate AI, evaluate how online dating has changed the way we meet people, or how machines and humans coexist in different areas of work.
This point of view is not new, the robotic he already raised, years ago, interest in the interactions between computers and humans. However, these studies have been carried out from different disciplineslike psychology or communication. The intention of this new discipline is create a better defined study framework you can focus on the analysis of the machines and their behavior.
