Machine behavior is an emerging field of study that explores the ways that intelligent agents behave, similarly to the way that many scientific disciplines explore the behavior of humans.
Artificial general intelligence (AGI, or strong AI) requires the ability of AI systems to find solutions to unfamiliar problems—situations that they have not been programmed for. Because advanced AI systems have some degree of autonomy and the capacity for independent action, their behavior is relatively unpredictable – it cannot be fully determined by examining their software any more than human behavior can be precisely predicted by examining our brains.
That unpredictability will only be compounded as AI systems become more intelligent and autonomous, and as they increasingly interact with people and each other in environments with varying degrees of human oversight. The study of machine behavior is recommended to catch unforeseen complications of AI autonomy, so that it can be dealt with before it leads to undesirable consequences.
According to Iyad Rahwan and Manuel Cebrian, writing in Nautilus, “This new discipline is concerned with the scientific study of machines, not as engineering artifacts, but as a new class of actors with their unique behavioral patterns and ecology.”