Proefschrift

172 Table 1: Continued. Author (s) Key concepts Miller et al. (2017) Decision-making is defined as: ‘an entity is in a situation, receives information about that situation, and selects and then implements a course of action.’ (p. 390). For an artificial agent to engage in ethical decision-making it needs to develop ethical expertise and is capable of self-doubt which is a high standard to reach. Openness to self-doubt is interpreted as three criteria for a machine being capable of making ethical decisions: ‘1. A capacity to sense some aspects of the outside world 2. An implementation of a function of merit that quantifies the acceptability of the current situation 3. A capacity to reprogram itself in order to improve performance in future situations.’ (p. 393) These three criteria require that an artificial agent has a mechanism, like a heuristic algorithm to analyse its past decisions and prepare for future decisions. It is not necessary for a machine to be able to make ethical decisions that its decision-making is similar to that of a human, but it is necessary to delineate which characteristics both a human and a machine require in order to make ethical decisions. van der Vecht (2009) Adjustable autonomy is defined as: ‘dynamically dealing with external influences on the decision-making process based on internal motivations.’ The agent can change its state in reaction to other agents and its environment and it can be achieved to get dynamic coordination. An agent can adjust its autonomy relative to other by making influence control a dynamic process. The degree of autonomy of an agent dictates the delegation of aspects of decision-making between the agent and an external actor. The level of autonomy is lowest in the concept of executive autonomy, followed by planning autonomy, goal determination and finally norm autonomy, which is the highest degree of autonomy. Adjustable autonomy allows for switching between these autonomy levels and dictates if these levels are controlled by the agent or an external actor. It is also possible to have a pre-planned process for the transfer-of-control in the decision-making process between an agent and a human. This can be used as a strategy for action or to change the coordination constraints, for example to request more time to reach a decision. ‘The definition of autonomy that is chosen can have implications for the decision-making process, both on single-agent and multi-agent decision-making’ (p. 19-20). APPENDIX C

RkJQdWJsaXNoZXIy MjY0ODMw