Orthogonality-Based Disentanglement of Responsibilities for Ethical Intelligent Systems

Promises in the Context of Humanoid Robot Morality

“…Combat robots: Work on combat robots which may not have lethal weapons on board can be found in Elands, et al [25], Aliman & Kester [26], and Aliman, et al [27]. A major risk of such weapons it that ware gets aut of control if these become operational in large numbers while fighting one-another rather than human opponents.…”

“…A major risk of such weapons it that ware gets aut of control if these become operational in large numbers while fighting one-another rather than human opponents. Aliman, et al [27], Sayler [32].…”

Hybrid Cognitive-Affective Strategies for AI Safety

“…Legal parameters but also legal norms and rules to limit the action space should be integrated in this process. In order to craft such societal-level augmented utility functions (also called ethical goal functions [16]), society would have to integrate scientific insights and facilitate the experience of counterfactual scenarios assisted for instance by VR and AR technology. Considering each cluster of instantiated dyadic cognitive templates and each perceiver, an algorithm could assimilate the corresponding relevant human-defined parameters with the human-defined weights and calculate the cardinal context-sensitive utility of the given scenario on which artificial intelligent systems could maximize.…”

“…For the governance of artificial intelligent systems which is a field of interest within both AI safety and AI ethics at an international level [365], it becomes crucial to design an appropriate goal specification framework able to encode the ethical and legal requirements within a given societal context. In this regard, different solutions have been proposed ranging from rule-based frameworks to methods based on updatable context-sensitive and perceiver-dependent ethical utility functions formulated at the societal level called ethical goal functions [16] (see Chapter 5). However, for any case in a given domain in which a society is supposed to contribute to implement a framework of AI governance, a process of ethical self-assessment attempting to unambiguously provide answers to the question on what society wants arises as necessity.…”

“…For this purpose, the utilization of VR or AR has been suggested for the experience of counterfactual preemptive simulations [14] in which people can scientifically calibrate their beliefs and as a result their moral judgements due to discernable differences in experience 1 . Ways in which the described ethical debiasing via VR/AR experiences could take place in the future and how this could be linked to a dynamical enhancement within a socio-technological feedback-loop [16] in the spirit of Positive Computing [93] are elucidated in Section 7.3.…”