Agent Morality via Counterfactuals in Logic Programming

Handbuch Maschinenethik, 2019

This chapter investigates the appropriateness of Logic Programming-based reasoning to machine ethics, an interdisciplinary field of inquiry that emerges from the need of imbuing autonomous agents with the capacity for moral decision making. The first part of the chapter aims at identifying morality viewpoints, as studied in moral philosophy and psychology, which are amenable to computational modeling, and then mapping them to appropriate Logic Programming-based reasoning features. The identified viewpoints are covered by two morality themes: moral permissibility and the dual-process model. In the second part, various Logic Programming-based reasoning features are applied to model these identified morality viewpoints, via classic moral examples taken off-the-shelf from the literature. For this purpose, our qualm system mainly employs a combination of the Logic Programming features of abduction, updating, and counterfactuals. These features are all supported jointly by Logic Programmi...

Electronic Proceedings in Theoretical Computer Science

Transparency is a key requirement for ethical machines. Verified ethical behavior is not enough to establish justified trust in autonomous intelligent agents: it needs to be supported by the ability to explain decisions. Logic Programming (LP) has a great potential for developing such perspective ethical systems, as in fact logic rules are easily comprehensible by humans. Furthermore, LP is able to model causality, which is crucial for ethical decision making.

Model-Based Reasoning in Science and Technology, 2016

Counterfactuals are conjectures about what would have happened, had an alternative event occurred. It provides lessons for the future by virtue of contemplating alternatives; it permits thought debugging; it supports a justification why different alternatives would have been worse or not better. Typical expressions are: "If only I were taller …", "I could have been a winner …", "I would have passed, were it not for …", "Even if ... the same would follow". Counterfactuals have been well studied in Linguistics, Philosophy, Physics, Ethics, Psychology, Anthropology, and Computation, but not much within Critical Thinking. The purpose of this study is to illustrate counterfactual thinking, through logic program abduction and updating, and inspired by Pearl's structural theory of counterfactuals, with an original application to morality, a common concern for critical thinking. In summary, we show counterfactual reasoning to be quite useful for critical thinking, namely about moral issues. Keywords Critical thinking Á Counterfactual reasoning Á Abduction Á Morality 1 Counterfactual Reasoning Counterfactual literally means contrary to the facts. Counterfactual reasoning involves thoughts on what could have happened, had some matter-action, outcome, etc.-been different in the past. Counterfactual thinking covers everyday experiences, like regret: "If only I had told her I love her!", "I should have studied harder"; or guilt responsibility, blame, causation: "If only I had said something

2007

Ethics is the science of duty, i.e. the science that elucidates the rules of the right behavior. Nevertheless, it seems that the way we rule our lives is intuitive and based on common sense. For instance, it is common to say that ethical rules are default rules, which means that they tolerate exceptions. Some authors argue that moral can only be grounded on particular cases while others defend the existence of general principles related to ethical rules. Our purpose here is not to justify the first or the second position, but to try to model ethical systems using artificial intelligence formalisms. More precisely, this is an attempt to show that progress in non-monotonic logics , which simulate common sense reasoning, provides a way to formalize different ethical conceptions. From a technical point of view, the model developed here makes use of the Answer Set Programming (ASP) formalism. It is applied to compare different ethical systems with respect to their attitude towards lying ...

Short papers call, Local Procs. 14th Intl. Conf …, 2007

This paper shows how moral decisions can be drawn computationally by using ACORDA, a working implementation of prospective logic programming. ACORDA is employed to model moral dilemmas, as they are able to prospectively look ahead at the consequences of hypothetical moral judgments. With this knowledge of consequences, moral rules are then used to decide the appropriate moral judgments. The whole moral reasoning is achieved via a priori constraints and a posteriori preferences on abductive stable models, two features available in ACORDA.

Ethics and Information Technology, 2005

A principal goal of the discipline of artificial morality is to design artificial agents to act as if they are moral agents. Intermediate goals of artificial morality are directed at building into AI systems sensitivity to the values, ethics, and legality of activities. The development of an effective foundation for the field of artificial morality involves exploring the technological and philosophical issues involved in making computers into explicit moral reasoners. The goal of this paper is to discuss strategies for implementing artificial morality and the differing criteria for success that are appropriate to different strategies.

2018

This paper makes use of high-level action languages to investigate aspects of causality that are central to ethical reasoning. We identify properties that causal relations assume and that determine how, as well as to what extent, we may ascribe ethical responsibility on their basis. The paper is structured in three parts. First, we present an extension of the Event Calculus that enables the agent to generate plans of actions, with the particularity that they integrate both actions and omissions. Second, we present an account of event-based causality that is grounded in the architecture of event preconditions and effects, and that distinguishes four types of causal relations contingent on the nature of the entities that compose them. Namely, it discriminates actions and omissions from automatic events, and produced outcomes from avoided ones. Third, we examine notions of scenario-based causality whose role it is to scrutinise and buttress the causal relations previously identified. I...

2021

We present an implementation of ethical reasoning using de- feasible logic. We evaluate this on two case studies illustrat-ing how changing information can affect the outcome of the ethical reasoning. We discuss the system’s strengths and weaknesses in terms of formal ethical reasoning, concluding that defeasible logic is a promising approach with particular value in the way it sepa- rates rules from exceptions but that additional work is needed, particularly in understanding how defeasible reasoning inter- acts with sequences/plans of action.

Advances in Intelligent Systems and Computing, 2020

This paper reviews the four fundamental frameworks available in normative ethics to underpin the most suitable strategy to facilitate the design of synthetic morality in the context of Highly Automated Systems (HAS). Based on research findings, it will present an updated multidimensional-scalar system of levels of automation specifically adapted to Highly Automated Systems (HAS) in the context of Human-Human-Interaction (HHI). This framework integrates the variables of autonomy, accountability, reparation, actions, contexts, access and inferences to build and facilitate the design of synthetic morality on highly automated unsupervised systems form a consequential perspective. As part of this process, a form of calculation emerges to facilitate the calibration of moral computational reasoning in the context of HAS.

1988

This paper shows that classical logic is inappropriate for hypothetical reasoning and develops an alternative logic for this purpose. The paper focuses on a form of hypothetical reasoning which appears computationally tractable. Specifically, Horn-clause logic is augmented with rules, called embedded implications, which can hypothetically add atomic formulas to a rulebase. By introducing the notion of rulebase independence, we show that these rules can express hypothetical queries which classical logic cannot; and by adopting methods from modal logic, we show these rules to be intuitionistic. In particular, they form a subset of intuitionistic logic having semantic properties similar to those of Horn-clause logic. This report is an expanded version of a paper published in the Proceedings of the Seventh National Conference on Artificial Intelligence, St. Paul, Minnesota, August 21--26 1988, American Association for Artificial Intelligence (AAAI). 1 Introduction Several researchers...