Implementations in Machine Ethics

Tolmeijer, Suzanne; Kneer, Markus; Sarasua, Cristina; Christen, Markus; Bernstein, Abraham

doi:10.1145/3419633

Cited by 93 publications

(71 citation statements)

References 126 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The aim of the current study was to identify ways in which to improve the fairness perceptions of AI interviews by examining the influences of three process design factors, namely (a) their positioning in the overall process, (b) applicants' sensitization to their potential to reduce human bias, and (c) human oversight of the AI decisionmaking process. The study thereby responded to the call for research on novel technologies for personnel selection (e.g., Blacksmith et al, 2016), as well as to the call for domain-specific work on the implementation of fair AI (e.g., Tolmeijer et al, 2020).…”

Section: Discussionmentioning

confidence: 99%

How to Improve Fairness Perceptions of AI in Hiring: The Crucial Role of Positioning and Sensitization

Hunkenschroer¹

2021

AIEJ

View full text Add to dashboard Cite

Companies increasingly deploy artificial intelligence (AI) technologies in their personnel recruiting and selection processes to streamline them, thus making them more efficient, consistent, and less human biased (Chamorro-Premuzic, Polli, & Dattner, 2019) . However, prior research found that applicants prefer face-to-face interviews compared with AI interviews, perceiving them as less fair (e.g., Acikgoz, Davison, Compagnone, & Laske, 2020) . Additionally, emerging evidence exists that contextual influences, such as the type of task for which AI is used (Lee, 2018) , or applicants’ individual differences (Langer, König, Sanchez, & Samadi, 2019) , may influence applicants’ reactions to AI-powered selection. The purpose of our study was to investigate whether adjusting process design factors may help to improve people's fairness perceptions of AI interviews. The results of our 2 x 2 x 2 online study (N = 404) showed that the positioning of the AI interview in the overall selection process, as well as participants’ sensitization to its potential to reduce human bias in the selection process have a significant effect on people’s perceptions of fairness. Additionally, these two process design factors had an indirect effect on overall organizational attractiveness mediated through applicants’ fairness perceptions. The findings may help organizations to optimize their deployment of AI in selection processes to improve people’s perceptions of fairness and thus attract top talent.

show abstract

Section: Discussionmentioning

confidence: 99%

How to Improve Fairness Perceptions of AI in Hiring: The Crucial Role of Positioning and Sensitization

Hunkenschroer¹

2021

AIEJ

View full text Add to dashboard Cite

show abstract

“…The AI literature on value alignment is typically divided between top-down, bottom-up, and hybrid approaches, as surveyed in Allen et al (2005), Tolmeijer et al (2021. In brief, top-down approaches focus on formalising ethical knowledge to encode it directly into the agent's behaviour, whereas bottom-up approaches resort on the agent learning the ethical knowledge by itself.…”

Section: Related Workmentioning

confidence: 99%

“…These approaches are convenient because they relieve the agent designer from the burden of defining the expected ethical behaviour of the agent for every possible situation. However, these approaches also suffer from well-known shortcomings, as discussed in Arnold et al (2017), Tolmeijer et al (2021), Gabriel (2020): (1) observing (learning from) human behaviour may ensure alignment with human habits but does not guarantee the learnt behaviour to be ethical; (2) the knowledge acquired by the agent through learning condenses experience in a manner that lacks of explicit representation (and reasoning) of the moral considerations that need to be taken into account (such as moral norms).…”

Section: Introductionmentioning

confidence: 99%

Instilling moral value alignment by means of multi-objective reinforcement learning

et al. 2022

View full text Add to dashboard Cite

AI research is being challenged with ensuring that autonomous agents learn to behave ethically, namely in alignment with moral values. Here, we propose a novel way of tackling the value alignment problem as a two-step process. The first step consists on formalising moral values and value aligned behaviour based on philosophical foundations. Our formalisation is compatible with the framework of (Multi-Objective) Reinforcement Learning, to ease the handling of an agent’s individual and ethical objectives. The second step consists in designing an environment wherein an agent learns to behave ethically while pursuing its individual objective. We leverage on our theoretical results to introduce an algorithm that automates our two-step approach. In the cases where value-aligned behaviour is possible, our algorithm produces a learning environment for the agent wherein it will learn a value-aligned behaviour.

show abstract

“…The aforementioned difficulties have led many machine ethicists to avoid virtue ethics completely, others to argue that it is inferior to other approaches, or that it is simply incomputable due to its uncodifiability (Arkin 2007;Bauer 2020;Tolmeijer et al 2020). But we will argue that there is a path for artificial virtue, provided that we give up on the project of trying to fully accommodate the theory's anthropocentric foundations.…”

Section: The Uncodifiability Of Virtuous Languagementioning

confidence: 99%

“…This moral behavior may be based on ethical theories such as teleological ethics, deontology, and virtue ethics, but not necessarily" (p. 506). 3 In a comprehensive survey of more than 50 implementations in machine ethics, not a single one had virtue ethics as its main focus (Tolmeijer et al 2020). Overall, only a handful of published articles explicitly deal with the construction of artificial virtuous agents (AVAs), and none of them provide any implementation details.…”

Section: Introductionmentioning

confidence: 99%

Artificial virtuous agents: from theory to machine implementation

Stenseke

2021

AI & Soc

View full text Add to dashboard Cite

Virtue ethics has many times been suggested as a promising recipe for the construction of artificial moral agents due to its emphasis on moral character and learning. However, given the complex nature of the theory, hardly any work has de facto attempted to implement the core tenets of virtue ethics in moral machines. The main goal of this paper is to demonstrate how virtue ethics can be taken all the way from theory to machine implementation. To achieve this goal, we critically explore the possibilities and challenges for virtue ethics from a computational perspective. Drawing on previous conceptual and technical work, we outline a version of artificial virtue based on moral functionalism, connectionist bottom–up learning, and eudaimonic reward. We then describe how core features of the outlined theory can be interpreted in terms of functionality, which in turn informs the design of components necessary for virtuous cognition. Finally, we present a comprehensive framework for the technical development of artificial virtuous agents and discuss how they can be implemented in moral environments.

show abstract

Implementations in Machine Ethics

Cited by 93 publications

References 126 publications

How to Improve Fairness Perceptions of AI in Hiring: The Crucial Role of Positioning and Sensitization

How to Improve Fairness Perceptions of AI in Hiring: The Crucial Role of Positioning and Sensitization

Instilling moral value alignment by means of multi-objective reinforcement learning

Artificial virtuous agents: from theory to machine implementation

Contact Info

Product

Resources

About