Journal of responsible technology最新文献

This study aims to map digitally mediated injustice and to understand how judicial versus non-judicial bodies contextualize and translate such harm into human rights violations. This study surveys judicial and quasi-judicial cases and case reports by non-judicial bodies, mainly civil society organizations, international organizations, and media. It divides digitally mediated harms identified through the survey into three categories: direct, structural, and hybrid harm. It then examines how these three forms of harm are represented and articulated in judicial judgments and case reports. To differentiate between the three forms of digitally mediated harm, the study uses Iris Young's political philosophy of structural injustice and Johan Galtung's account of structural violence in peace studies. The focus of this study is understanding the forms of injustices that are present but rendered invisible because of how they are contextualized. Therefore, the epistemology of absence is applied as the theoretical approach, that is, interpretation of absence and invisibility. The epistemology of absence facilitates the identification of structural and intersectional injustices that are not articulated in the same way they are experienced in society. The assessment reveals four observations. (1) Structural injustice is rarely examined through a conventional adjudicatory process. (2) Harms of structural quality examined by courts are narrowly interpreted when translated into rights violations. (3) The right to privacy, often presented as a gateway right, addresses structural injustice only partially, as this right has a subject-centric narrow interpretation currently. (4) There are limitations to the mainstream way of seeing and representing risks and injustices in the digital space, and such a view yields metonymic reasoning when framing digitally produced harms. As a result, the conventional way of contextualization is blind to unconventional experiences of vulnerability, which renders structural and intersectional injustices experienced by marginalized communities invisible.

As AI-based models take an increasingly central role in our lives, so does the concern for fairness. In recent years, mounting evidence reveals how vulnerable AI models are to bias and the challenges involved in detection and mitigation. Our contribution is three-fold. Firstly, we gather name disparity tables across protected groups, allowing us to estimate sensitive attributes (gender, race). Using these estimates, we compute bias metrics given a classification model’s predictions. We leverage only names/zip codes; hence, our method is model and feature agnostic. Secondly, we offer an open-source Python package that produces a bias detection report based on our method. Finally, we demonstrate that names of older individuals are better predictors of race and gender and that double surnames are a reasonable predictor of gender. We tested our method on publicly available datasets (US Congress) and classifiers (COMPAS) and found it to be consistent with them.

The AI Ethics literature, aimed to responsibly develop AI systems, widely agrees on the fact that society is in dire need for effective accountability mechanisms with regards to AI systems. Particularly, machine learning (ML) systems cause reason for concern due to their opaque and self-learning characteristics. Nevertheless, what such accountability mechanisms should look like remains either largely unspecified (e.g., ‘stakeholder input’) or ineffective (e.g., ‘ethical guidelines’). In this paper, I argue that the difficulty to formulate and develop effective accountability mechanisms lies partly in the predominant focus on Mill's harm's principle, rooted in the conception of freedom as non-interference. A strong focus on harm overcasts other moral wrongs, such as potentially problematic power dynamics between those who shape the system and those affected by it. I propose that the neo-republican conception of freedom as non-domination provides a suitable framework to inform responsible ML development. Domination, understood by neo-republicans, is a moral wrong as it undermines the potential for human flourishing. In order to mitigate domination, neo-republicans plead for accountability mechanisms that minimize arbitrary relations of power. Neo-republicanism should hence inform responsible ML development as it provides substantive and concrete grounds when accountability mechanisms are effective (i.e. when they are non-dominating).

Accountability encompasses multiple aspects such as responsibility, justification, reporting, traceability, audit, and redress so as to satisfy diverse requirements of different stakeholders— consumers, regulators, developers, etc. In order to take into account needs of different stakeholders and thus, to put into practice accountability in Artificial Intelligence, the notion of empathy can be quite effective. Empathy is the ability to be sensitive to the needs of someone based on understanding their affective states and intentions, caring for their feelings, and socialization, which can help in addressing the social-technical challenges associated with accountability. The goal of this paper is twofold. First, we elucidate the connections between empathy and accountability, drawing find- ings from various disciplines like psychology, social science, and organizational science. Second, we suggest potential pathways to incorporate empathy.

Billions of people interact within platform-based ecosystems containing the personal data of their daily lives. Data which have become rigorously creatable, processable, and shareable. Here, platform providers facilitate interactions between three types of relevant actors: users, service providers, and third parties. Research in the information systems field has shown that platform providers influence their platform ecosystems to promote the contributions of service providers and exercise control by utilizing boundary resources. Through a socio-techno-legal analysis of two high-profile cases and their application on the General Data Protection Regulation (GDPR) we show that the boundary resource design, arrangement, and interplay can influence whether and to what extent platform providers are accountable for platform providers unlawful personal data processing in platform ecosystems. The findings can have a huge impact to account actors for personal data misusage in platform ecosystems and, thus, the protection of personal liberty and rights in such socio-technical systems.

Civilian Unmanned Aerial Vehicle (UAV) is an emerging technology in Indian crop insurance applications. The technology is new to an agro-based country like India with diverse socio-cultural norms and values. However, in such a diverse democracy, UAV governance and deployment pose a significant challenge and risk. In other words, charting out a proper framework for a risk-free implementation of this governance has emerged as a leading research topic in the concerned discipline. In innovations literature, Responsible Innovation (RI) takes care of emerging technology governance; thus, RI becomes significant as a theoretical framework. The study is intended to find out how the framework of RI enables responsible governance and also who are the main actors and stakeholders of governance and deployment of civilian UAVs in crop insurance applications in India? An in-depth interview method and snowball sampling technique have been employed to identify interviewees from Delhi, Gujarat, and Rajasthan. Findings suggest that civilian UAV is effective in handling risk, crop damage assessment, and claim settlement. The RI approach, through its dimensions and steps, enables equal participation and deliberation among all the actors and stakeholders of UAV governance that consists of government bodies, research organizations, insurance agencies, local administration, and farmers. Effective regulations, adhering to accountability, and responsibility promote responsible governance.

AI as decision support supposedly helps human agents make ‘better’ decisions more efficiently. However, research shows that it can, sometimes greatly, influence the decisions of its human users. While there has been a fair amount of research on intended AI influence, there seem to be great gaps within both theoretical and practical studies concerning unintended AI influence. In this paper I aim to address some of these gaps, and hope to shed some light on the ethical and moral concerns that arise with unintended AI influence. I argue that unintended AI influence has important implications for the way we perceive and evaluate human-AI interaction. To make this point approachable from both the theoretical and practical side, and to avoid anthropocentrically-laden ambiguities, I introduce the notion of decision points. Based on this, the main argument of this paper will be presented in two consecutive steps: i) unintended AI influence doesn’t allow for an appropriate determination of decision points - this will be introduced as decision-point-dilemma, and ii) this has important implications for the ascription of responsibility.

Algorithmic decision support systems are widely applied in domains ranging from healthcare to journalism. To ensure that these systems are fair and accountable, it is essential that humans can maintain meaningful agency, understand and oversee algorithmic processes. Explainability is often seen as a promising mechanism for enabling human-in-the-loop, however, current approaches are ineffective and can lead to various biases. We argue that explainability should be tailored to support naturalistic decision-making and sensemaking strategies employed by domain experts and novices. Based on cognitive psychology and human factors literature review we map potential decision-making strategies dependant on expertise, risk and time dynamics and propose the conceptual Expertise, Risk and Time Explainability framework, intended to be used as explainability design guidelines. Finally, we present a worked example in journalism to illustrate the applicability of our framework in practice.

Algorithmic decision support is increasingly used in a whole array of different contexts and structures in various areas of society, influencing many people’s lives. Its use raises questions, among others, about accountability, transparency and responsibility. While there is substantial research on the issue of algorithmic systems and responsibility in general, there is little to no prior research on organisational responsibility and its attribution. Our article aims to fill that gap; we give a brief overview of the central issues connected to ADS, responsibility and decision-making in organisational contexts and identify open questions and research gaps. Furthermore, we describe a set of guidelines and a complementary digital tool to assist practitioners in mapping responsibility when introducing ADS within their organisational context.