RESUMO
Advances in artificial intelligence (AI) raise important questions about whether people view moral evaluations by AI systems similarly to human-generated moral evaluations. We conducted a modified Moral Turing Test (m-MTT), inspired by Allen et al. (Exp Theor Artif Intell 352:24-28, 2004) proposal, by asking people to distinguish real human moral evaluations from those made by a popular advanced AI language model: GPT-4. A representative sample of 299 U.S. adults first rated the quality of moral evaluations when blinded to their source. Remarkably, they rated the AI's moral reasoning as superior in quality to humans' along almost all dimensions, including virtuousness, intelligence, and trustworthiness, consistent with passing what Allen and colleagues call the comparative MTT. Next, when tasked with identifying the source of each evaluation (human or computer), people performed significantly above chance levels. Although the AI did not pass this test, this was not because of its inferior moral reasoning but, potentially, its perceived superiority, among other possible explanations. The emergence of language models capable of producing moral responses perceived as superior in quality to humans' raises concerns that people may uncritically accept potentially harmful moral guidance from AI. This possibility highlights the need for safeguards around generative language models in matters of morality.
Assuntos
Inteligência Artificial , Princípios Morais , Humanos , Inteligência Artificial/ética , Feminino , Masculino , Adulto , Adulto Jovem , Pessoa de Meia-Idade , JulgamentoRESUMO
Scholars have proposed that incarceration rates might be reduced by a requirement that judges justify incarceration decisions with respect to their operational costs (e.g., prison capacity). In an Internet-based vignette experiment (N = 214), we tested this prediction by examining whether criminal punishment judgments (prison vs. probation) among university undergraduates would be influenced by a prompt to provide a justification for one's judgment, and by a brief message describing prison capacity costs. We found that (1) the justification prompt alone was sufficient to reduce incarceration rates, (2) the prison capacity message also independently reduced incarceration rates, and (3) incarceration rates were most strongly reduced (by about 25%) when decision makers were asked to justify their sentences with respect to the expected capacity costs. These effects survived a test of robustness and occurred regardless of whether participants reported that prison costs should influence judgments of incarceration. At the individual crime level, the least serious crimes were most amenable to reconsideration for probation. These findings are important for policymakers attempting to manage high incarceration rates.
RESUMO
To function during social interactions, we must be able to consider and coordinate our actions with other people's perspectives. This process unfolds from decision-making, to anticipation of that decision's consequences, to feedback about those consequences, in what can be described as a "cascade" of three phases. The iterated Prisoner's Dilemma (iPD) task, an economic-exchange game used to illustrate how people achieve stable cooperation over repeated interactions, provides a framework for examining this "social decision cascade". In the present study, we examined neural activity associated with the three phases of the cascade, which can be isolated during iPD game rounds. While undergoing functional magnetic resonance imaging (fMRI), 31 adult participants made a) decisions about whether to cooperate with a co-player for a monetary reward, b) anticipated the co-player's decision, and then c) learned the co-player's decision. Across all three phases, participants recruited the temporoparietal junction (TPJ) and the dorsomedial prefrontal cortex (dmPFC), regions implicated in numerous facets of social reasoning such as perspective-taking and the judgement of intentions. Additionally, a common distributed neural network underlies both decision-making and feedback appraisal; however, differences were identified in the magnitude of recruitment between both phases. Furthermore, there was limited evidence that anticipation following the decision to defect evoked a neural signature that is distinct from the signature of anticipation following the decision to cooperate. This study is the first to delineate the neural substrates of the entire social decision cascade in the context of the iPD game.
Assuntos
Comportamento Cooperativo , Tomada de Decisões/fisiologia , Rede Nervosa/diagnóstico por imagem , Córtex Pré-Frontal/diagnóstico por imagem , Recompensa , Adolescente , Adulto , Feminino , Humanos , Imageamento por Ressonância Magnética , Masculino , Dilema do Prisioneiro , Adulto JovemRESUMO
BACKGROUND: Social anxiety is characterized by a tendency to overestimate the likelihood of negative outcomes and consequences before, during, and after interpersonal interactions with social partners. Recent evidence suggests that a network of brain regions critical for perspective-taking, threat appraisal, and uncertainty resolution may function atypically in those prone to social anxiety. In this study, we used functional magnetic resonance imaging to examine neural activity in specific regions of interest in a sample of young adults who endorsed high or low levels of social anxiety. METHODS: We recruited 31 college student volunteers (age: 18-28 years), categorized as having high or low anxiety based on their Liebowitz Social Anxiety Scale-Self Report scores. These participants were each scanned while playing the iterated Prisoner's Dilemma game with three computerized confederates, two of whom they were deceived to believe were human co-players. This study focuses on data collected during play with the presumed humans. Regions of interest were defined for the temporoparietal junction, anterior midcingulate, and dorsomedial prefrontal cortex. Average weighted mean blood-oxygen-level-dependent signals for each subject were extracted and analyzed using mixed design analyses of variance to detect group differences in activation during decision-making, anticipation, and appraisal of round outcomes during the game. RESULTS: Behavior analysis revealed that the high-anxiety group was more likely to defect than the low-anxiety group. Neuroimaging analysis showed that the high-anxiety group exhibited elevated blood-oxygen-level-dependent activity relative to the low-anxiety group in all three regions during the social feedback appraisal phase but not during decision-making or the anticipation of interaction outcomes. CONCLUSIONS: These findings provide evidence that some behaviors linked to cognitive biases associated with social anxiety may be mediated by a network of regions involved in recognizing and processing directed social information. Future investigation of the neural basis of cognition and bias in social anxiety using the prisoner's dilemma and other economic-exchange tasks is warranted. These tasks appear to be highly effective, functional magnetic resonance imaging-compatible methods of probing altered cognition and behavior associated with anxiety and related conditions.
RESUMO
In recent years, a number of prominent scientists have argued that free will is an illusion, appealing to evidence demonstrating that information about brain activity can be used to predict behavior before people are aware of having made a decision. These scientists claim that the possibility of perfect prediction based on neural information challenges the ordinary understanding of free will. In this paper we provide evidence suggesting that most people do not view the possibility of neuro-prediction as a threat to free will unless it also raises concerns about manipulation of the agent's behavior. In Experiment 1 two scenarios described future brain imaging technology that allows perfect prediction of decisions and actions based on earlier neural activity, and this possibility did not undermine most people's attributions of free will or responsibility, except in the scenario that also allowed manipulation. In Experiment 2 the scenarios increased the salience of the physicalist implications of neuro-prediction, while in Experiment 3 the scenarios suggested dualism, with perfect prediction by mindreaders. The patterns of results for these two experiments were similar to the results in Experiment 1, suggesting that participants do not understand free will to require specific metaphysical conditions regarding the mind-body relation. Most people seem to understand free will in a way that is not threatened by perfect prediction based on neural information, suggesting that they believe that just because "my brain made me do it," that does not mean that I didn't do it of my own free will.
Assuntos
Encéfalo/fisiologia , Intuição/fisiologia , Neurociências , Volição/fisiologia , Atenção , Feminino , Humanos , Masculino , Autonomia Pessoal , Adulto JovemRESUMO
In this paper, we present the results of the construction and validation of a new psychometric tool for measuring beliefs about free will and related concepts: The Free Will Inventory (FWI). In its final form, FWI is a 29-item instrument with two parts. Part 1 consists of three 5-item subscales designed to measure strength of belief in free will, determinism, and dualism. Part 2 consists of a series of fourteen statements designed to further explore the complex network of people's associated beliefs and attitudes about free will, determinism, choice, the soul, predictability, responsibility, and punishment. Having presented the construction and validation of FWI, we discuss several ways that it could be used in future research, highlight some as yet unanswered questions that are ripe for interdisciplinary investigation, and encourage researchers to join us in our efforts to answer these questions.
Assuntos
Atitude , Autonomia Pessoal , Punição , Responsabilidade Social , Adolescente , Adulto , Idoso , Idoso de 80 Anos ou mais , Análise Fatorial , Feminino , Humanos , Masculino , Pessoa de Meia-Idade , Psicometria/instrumentação , Reprodutibilidade dos Testes , Inquéritos e Questionários , Adulto JovemRESUMO
Free will is a set of capacities for conscious choice and control of actions and is essential for moral responsibility. While determinism is traditionally discussed as the main potential challenge to free will and responsibility, other potential challenges exist and need to be considered by philosophers and scientists. The cognitive sciences are relevant to free will both to study how people understand free will and potential challenges to it, and to study whether these challenges are supported by relevant scientific evidence. WIREs Cogn Sci 2012, 3:439-449. doi: 10.1002/wcs.1181 For further resources related to this article, please visit the WIREs website.
RESUMO
In this paper, we examine Adam Feltz and Edward Cokely's recent claim that "the personality trait extraversion predicts people's intuitions about the relationship of determinism to free will and moral responsibility" (INSERT REFERENCE). We will first present some criticisms of their work before briefly examining the results of a recent study of our own. We argue that while Feltz and Cokely have their finger on the pulse of an interesting and important issue, they have not established a robust and stable connection between extraversion and compatibilist-friendly intuitions.
Assuntos
Intuição , Temperamento , Extroversão Psicológica , Humanos , Responsabilidade SocialRESUMO
Daniel Wegner argues that conscious will is an illusion. I examine the adequacy of his theory of apparent mental causation and whether, if accurate, it suggests that our experience of agency and authorship should be considered illusory. I examine various interpretations of this claim and raise problems for each interpretation. I also distinguish between the experiences of agency and authorship.