As we are living in an era of increased development of Artificial Intelligence (AI), newethical dilemmas arise. Some countries, states, technicians, developers of AI and researchersin different faculties have already observed that AI could get out of control and affectsocieties and human beings in a disastrous way. This thesis casts some light on the importanceof investigating one key concept in AI, “Autonomy”, and how different understandings of thedefinition have no effects on the ethical consequences. Immanuel Kant’s Moral Philosophyand Martha C. Nussbaums model of how to create capabilities for humans is the ethicaltheories used to evaluate different views on the concept autonomy. The first perspective onautonomy that is evaluated is Paula Boddington’s view that a definition of autonomy is notneeded but ethical codes for AI should be developed instead. The second view that isevaluated is the confusion about autonomy in machines and humans, and different ways ofunderstanding functional autonomy. This thesis argues for the importance of defining theconcept of autonomy for the benefit of mankind in working together for a benevolent AI. Aquestion that arises from what Kant calls “the Divine Spark” that all humans have and thatmake her worthy of respects; is it possible to put the “Divine Spark” into an AI?