Proefschrift

85 3 CONCEPTUAL FRAMEWORK and not be viewed as a monolithic concept. They suggest four types of feedback: complaint, suggestions, monitoring and satisfaction. These types do not fit the case of Autonomous Weapon Systems, because these are based on user or customer feedback. Also, the more feedback you seek the more capacity it takes to respond and act on it. Therefore, the type of feedback should be limited. In case of the Comprehensive Human Oversight Framework the relevant feedback is information on compliance or non-compliance of the system with the criteria that were set pre-deployment. These criteria should be set during the interpretation stage and can be done during drafting of the norms after the value deliberation. The frequency of the feedback is dependent on the type of accountability process that is followed: ͳ Political accountability process will often be conducted as part of the post mission review process or after an incident that requires a political hearing; ͳ Legal accountability process will be conducted after a violation of the law has occurred; ͳ Administrative accountability process will be performed on request of an institution or actor; ͳ Professional account process will be conducted when internal regulation or codes of conduct are violated. In the case of deployment of an Autonomous Weapon System this could be when during the After Action Review a deviation of the pre-determined criteria is observed. Based on the type of accountability, it is not possible to set a specific predetermined time frame for the feedback process as it is dependent on the type of accountability process described above. Nevertheless, when an article 36 weapon review process can be standardized as suggested above a frequency can be determined. The article 36 review process is mostly conducted when new weapons and methods or means of warfare are studied, developed, acquired or adopted. In the case of deployment of Autonomous Weapon Systems this frequency needs to be increased due to the autonomous decision-making of AI. Determining the nature of the frequency should be part of the design of the standardized article 36 review process. 4. Tools In selecting the type of tools – no, low or high tech – (see Table 7) for conducting the feedback process two criteria are important; 1) expanding reach by levering new technologies and 2) ensuring inclusivity of participation in order not to

RkJQdWJsaXNoZXIy MjY0ODMw