Ensuring the trustworthiness of autonomous systems and artificial intelligence is an important interdisciplinary endeavour. In this position paper, we argue that this endeavour will benefit from technical advancements in capturing various forms of responsibility, and we present a comprehensive research agenda to achieve this. In particular, we argue that ensuring the reliability of autonomous system can take advantage of technical approaches for quantifying degrees of responsibility and for coordinating tasks based on that. Moreover, we deem that, in certifying the legality (...) of an AI system, formal and computationally implementable notions of _responsibility_, _blame_, _accountability_, and _liability_ are applicable for addressing potential responsibility gaps (i.e. situations in which a group is responsible, but individuals’ responsibility may be unclear). This is a call to enable AI systems themselves, as well as those involved in the design, monitoring, and governance of AI systems, to represent and reason about who can be seen as responsible in prospect (e.g. for completing a task in future) and who can be seen as responsible retrospectively (e.g. for a failure that has already occurred). To that end, in this work, we show that across all stages of the design, development, and deployment of trustworthy autonomous systems (TAS), responsibility reasoning should play a key role. This position paper is the first step towards establishing a road map and research agenda on how the notion of responsibility can provide novel solution concepts for ensuring the _reliability_ and _legality_ of TAS and, as a result, enables an effective embedding of AI technologies into society. (shrink)

Since the 1960s, scientists, engineers, and healthcare professionals have developed brain–computer interface (BCI) technologies, connecting the user’s brain activity to communication or motor devices. This new technology has also captured the imagination of publics, industry, and ethicists. Academic ethics has highlighted the ethical challenges of BCIs, although these conclusions often rely on speculative or conceptual methods rather than empirical evidence or public engagement. From a social science or empirical ethics perspective, this tendency could be considered problematic and even technocratic because (...) of its disconnect from publics. In response, our trinational survey (Germany, Canada, and Spain) reports public attitudes toward BCIs (N = 1,403) on ethical issues that were carefully derived from academic ethics literature. The results show moderately high levels of concern toward agent-related issues (e.g., changing the user’s self) and consequence-related issues (e.g., new forms of hacking). Both facets of concern were higher among respondents who reported as female or as religious, while education, age, own and peer disability, and country of residence were associated with either agent-related or consequence-related concerns. These findings provide a first look at BCI attitudes across three national contexts, suggesting that the language and content of academic BCI ethics may resonate with some publics and their values. (shrink)

Bioethics has made a compelling case for the role of experience and empirical research in ethics. This may explain why the movement for empirical ethics has such a firm grounding in bioethics. However, the theoretical framework according to which empirical research contributes to ethics—and the specific role it can or should play—remains manifold and unclear. In this paper, we build from pragmatic theory stressing the importance of experience and outcomes in establishing the meaning of ethics concepts. We then propose three (...) methodological steps according to which the meaning of ethics concepts can be refined based on experience and empirical research: function identification, function enrichment, and function testing. These steps are explained and situated within the broader commitment of pragmatic ethics to a perspective of moral growth and human flourishing. We hope that this proposal will give specific direction to the bridging of theoretical and empirical research in ethics and thus support stronger actualization of ethics concepts. (shrink)

Autonomous vehicles —and accidents they are involved in—attest to the urgent need to consider the ethics of artificial intelligence. The question dominating the discussion so far has been whether we want AVs to behave in a ‘selfish’ or utilitarian manner. Rather than considering modeling self-driving cars on a single moral system like utilitarianism, one possible way to approach programming for AI would be to reflect recent work in neuroethics. The agent–deed–consequence model :3–20, 2014a, Behav Brain Sci 37:487–488, 2014b) provides a (...) promising descriptive and normative account while also lending itself well to implementation in AI. The ADC model explains moral judgments by breaking them down into positive or negative intuitive evaluations of the agent, deed, and consequence in any given situation. These intuitive evaluations combine to produce a positive or negative judgment of moral acceptability. For example, the overall judgment of moral acceptability in a situation in which someone committed a deed that is judged as negative would be mitigated if the agent had good intentions and the action had a good consequence. This explains the considerable flexibility and stability of human moral judgment that has yet to be replicated in AI. This paper examines the advantages and disadvantages of implementing the ADC model and how the model could inform future work on ethics of AI in general. (shrink)

Empirical neuroethics models have always had normative ambitions. Older models attempted to debunk traditional moral theories, whereas newer models attempt to fit their empirical and normative claims with them. The issue of normative significance as it pertains to the use of social science methodology on moral intuitions remains open. This paper analyzes the Is/Ought gap and the empirical underpinnings of influential constructivist approaches in order to argue that the normative ambitions of empirical neuroethics models are not necessarily always misguided. The (...) author clarifies how legitimate normative implications can be dissociated from overreaches and argues that moral intuitions could be used as the factual basis for ethics by employing the distinction between natural facts, social facts, and social constructs. Ultimately, morality can be fruitfully studied and informed by natural and social sciences and, thorough ethical reflection, ought also to inform empirical research. (shrink)

Recent research in empirical moral psychology attempts to understand (rather than place judgment on) the salient normative differences that laypeople have when making moral decisions by using survey methodology that is based on the operationalized principles from moral theories. The PPIMT is the first measure designed to assess respondents’ preference for the precepts implied in the three dominant moral theories: virtue ethics, deontology, and consequentialism. The current study used a latent modeling approach to determine the most theoretically and psychometrically-sound model (...) for the PPIMT using a combined sample of college students from a southeastern university in U.S. and MTurk respondents. The PPIMT model fit was acceptable (χ2 = 84.125, df = 40, p = 0.001; RMSEA = 0.052, 90%CI = 0.037 to 0.068; CFI = 0.980; SRMR = 0.035) with four items for Virtue, four items for Deontology, and three items for Consequentialism. (shrink)

The authors interpret the data to mean that patients think that their physicians should make relevant decisions in Learning Health System based trials, and label that as being of 'utmost importance'. However, the patients themselves (in the excerpts provided) emphasize that trust in physicians is instrumental for obtaining protection of patient's bests interests (which seems to be of utmost importance for patients). Furthermore, the perceived bias regarding outcome certainty deserves more discussion. Namely, the decision to defer to physician's opinion is (...) linked to perceived certainty of best clinical outcomes and instrumental for protection of best interest, while it lacks appreciation of the level of uncertainty that exists in clinical decision-making. Finally, a discussion on what are the implications of patient perspectives for individual, vs. relational autonomy is something the authors don't pursue, but instead use "physician-patient autonomy" throughout the paper. (shrink)

In their paper “Algorithms for Ethical Decision-Making in the Clinic: A Proof of Concept,” Meier et al. present the design and preliminary results of a proof-of-concept clinical ethics algor...

Research in empirical moral psychology has consistently found negative correlations between morality and both risk-taking, as well as psychopathic tendencies. However, prior research did not sufficiently explore intervening or moderating factors. Additionally, prior measures of moral preference have a pronounced lack of ecological validity. This study seeks to address these two gaps in the literature. First, this study used Preference for Precepts Implied in Moral Theories, which offers a novel, more nuanced and ecologically valid measure of moral judgment. Second, the (...) current study examined if risk taking moderates the relationships between psychopathic tendencies and moral judgment. Results indicated that models which incorporated risk-taking as a moderator between psychopathic tendencies and moral judgment were a better fit to the data than those that incorporated psychopathic tendencies and risk-taking as exogenous variables, suggesting that the association between psychopathic tendencies and moral judgment is influenced by level of risk-taking. Therefore, future research investigating linkages between psychopathic tendencies and moral precepts may do well to incorporate risk-taking and risky behaviors to further strengthen the understanding of moral judgment in these individuals. (shrink)

Stimulant drugs, transcranial magnetic stimulation, brain-computer interfaces, and even genetic modifications are all discussed as forms of potential cognitive enhancement. Cognitive enhancement can be conceived as a benefit-seeking strategy used by healthy individuals to enhance cognitive abilities such as learning, memory, attention, or vigilance. This phenomenon is hotly debated in the public, professional, and scientific literature. Many of the statements favoring cognitive enhancement (e.g., related to greater productivity and autonomy) or opposing it (e.g., related to health-risks and social expectations) rely (...) on claims about human welfare and human flourishing. But with real-world evidence from the social and psychological sciences often missing to support (or invalidate) these claims, the debate about cognitive enhancement is stalled. In this paper, we describe a set of crucial debated questions about psychological and social aspects of cognitive enhancement (e.g., intrinsic motivation, well-being) and explain why they are of fundamental importance to address in the cognitive enhancement debate and in future research. We propose studies targeting social and psychological outcomes associated with cognitive enhancers (e.g., stigmatization, burnout, mental well-being, work motivation). We also voice a call for scientific evidence, inclusive of but not limited to biological health outcomes, to thoroughly assess the impact of enhancement. This evidence is needed to engage in empirically informed policymaking, as well as to promote the mental and physical health of users and non-users of enhancement. (shrink)

What ethical principles should autonomous machines follow? How do we implement these principles, and how do we evaluate these implementations? These are some of the critical questions Vivek Nallur asks in his essay “Landscape of Machine Implemented Ethics (2020).” He provides a broad, insightful survey of answers to these questions, especially focused on the implementation question. In this commentary, I will first critically summarize the main themes and conclusions of Nallur’s essay and then expand upon the landscape that Nallur presents (...) by suggesting additional approaches to machine ethics. The approaches I discuss reflect normative ethical theories and need not be applied only to machines, although machine ethics is the focus here. The overall goal is to open up further questions and research possibilities as society searches for the best approach to machine ethics. (Due to an unfortunate miscommunication with the copy editor an important reference was omitted from this article. The reference that should be included is: Nallur, V. (2020). Landscape of Machine Implemented Ethics. Science and Engineering Ethics 26 (5):2381-2399.). (shrink)

In this dissertation, I construct scientifically and practically adequate moral analogs of cognitive heuristics and biases. Cognitive heuristics are reasoning “shortcuts” that are efficient but flawed. Such flaws yield systematic judgment errors—i.e., cognitive biases. For example, the availability heuristic infers an event’s probability by seeing how easy it is to recall similar events. Since dramatic events, such as airplane crashes, are disproportionately easy to recall, this heuristic explains systematic overestimations of their probability (availability bias). The research program on cognitive heuristics (...) and biases (e.g., Daniel Kahneman’s work) has been scientifically successful and has yielded useful error-prevention techniques—i.e., cognitive debiasing. I attempt to apply this framework to moral reasoning to yield moral heuristics and biases. For instance, a moral bias of unjustified differences in the treatment of particular animal species might be partially explained by a moral heuristic that dubiously infers animals’ moral status from their aesthetic features. While the basis for identifying judgments as cognitive errors is often unassailable (e.g., per violating laws of logic), identifying moral errors seemingly requires appealing to moral truth, which, I argue, is problematic within science. Such appeals can be avoided by repackaging moral theories as mere “standards-of-interest” (a la non-normative metrics of purportedly right-making features/properties). However, standards-of-interest do not provide authority, which is needed for effective debiasing. Nevertheless, since each person deems their own subjective morality authoritative, subjective morality (qua standard-of-interest and not moral subjectivism) satisfies both scientific and practical concerns. As such, (idealized) subjective morality grounds a moral analog of cognitive biases—namely, subjective moral biases (e.g., committed anti-racists unconsciously discriminating). I also argue that "cognitive heuristic" is defined by its contrast with rationality. Consequently, heuristics explain biases, which are also so defined. However, such contrasting with rationality is causally irrelevant to cognition. This frustrates the presumed usefulness of the kind, heuristic, in causal explanation. As such, in the moral case, I jettison the role of causal explanation and tailor categories solely for contrastive explanation. As such, “moral heuristic” is replaced with "subjective moral fallacy," which is defined by its contrast with subjective morality and explains subjective moral biases. The resultant subjective moral biases and fallacies framework can undergird future empirical research. (shrink)