Search

Artificial Intelligence (AI) can be harnessed to create sophisticated social and moral scoring systems – enabling people and organisations to form judgements of others at scale. While this capability has many useful applications – e.g., matching romantic partners who are aligned in their moral principles, it also raises many ethical questions. For example, there is widespread concern about the use of social credit systems in the political domain. In this project, we approach this topic from a psychological perspective. With experimental evidence, we show that the acceptability of moral scoring by AI depends on its perceived accuracy, and that perceived accuracy is compromised by people's tendency to see themselves as morally peculiar, and thus less characterizable by AI. That is, we suggest that people overestimate the peculiarity of their moral profile, believe that AI will neglect this peculiarity, and resist for this reason the introduction of moral scoring by AI.

Conversation with a trained chatbot can reduce conspiratorial beliefs.

Humans, aware of the social costs associated with false accusations, are generally hesitant to accuse others of lying. Our study shows how lie detection algorithms disrupt this social dynamic. We develop a supervised machine-learning classifier that surpasses human accuracy and conduct a large-scale incentivized experiment manipulating the availability of this lie-detection algorithm. In the absence of algorithmic support, people are reluctant to accuse others of lying, but when the algorithm becomes available, a minority actively seeks its prediction and consistently relies on it for accusations. Although those who request machine predictions are not inherently more prone to accuse, they more willingly follow predictions that suggest accusation than those who receive such predictions without actively seeking them., Competing Interests: The authors declare that they have no conflict of interest., (© 2024 The Authors.)

Generative artificial intelligence (AI) has the potential to both exacerbate and ameliorate existing socioeconomic inequalities. In this article, we provide a state-of-the-art interdisciplinary overview of the potential impacts of generative AI on (mis)information and three information-intensive domains: work, education, and healthcare. Our goal is to highlight how generative AI could worsen existing inequalities while illuminating how AI may help mitigate pervasive social problems. In the information domain, generative AI can democratize content creation and access but may dramatically expand the production and proliferation of misinformation. In the workplace , it can boost productivity and create new jobs, but the benefits will likely be distributed unevenly. In education , it offers personalized learning, but may widen the digital divide. In healthcare , it might improve diagnostics and accessibility, but could deepen pre-existing inequalities. In each section, we cover a specific topic, evaluate existing research, identify critical gaps, and recommend research directions, including explicit trade-offs that complicate the derivation of a priori hypotheses. We conclude with a section highlighting the role of policymaking to maximize generative AI's potential to reduce inequalities while mitigating its harmful effects. We discuss strengths and weaknesses of existing policy frameworks in the European Union, the United States, and the United Kingdom, observing that each fails to fully confront the socioeconomic challenges we have identified. We propose several concrete policies that could promote shared prosperity through the advancement of generative AI. This article emphasizes the need for interdisciplinary collaborations to understand and address the complex challenges of generative AI., (© The Author(s) 2024. Published by Oxford University Press on behalf of National Academy of Sciences.)

To succeed, we posit that research cartography will require high-throughput natural description to identify unknown unknowns in a particular design space. High-throughput natural description, the systematic collection and annotation of representative corpora of real-world stimuli, faces logistical challenges, but these can be overcome by solutions that are deployed in the later stages of integrative experiment design.

Moral psychology was shaped around three categories of agents and patients: humans, other animals, and supernatural beings. Rapid progress in artificial intelligence has introduced a fourth category for our moral psychology to deal with: intelligent machines. Machines can perform as moral agents, making decisions that affect the outcomes of human patients or solving moral dilemmas without human supervision. Machines can be perceived as moral patients, whose outcomes can be affected by human decisions, with important consequences for human-machine cooperation. Machines can be moral proxies that human agents and patients send as their delegates to moral interactions or use as a disguise in these interactions. Here we review the experimental literature on machines as moral agents, moral patients, and moral proxies, with a focus on recent findings and the open questions that they suggest.

The ability of humans to create and disseminate culture is often credited as the single most important factor of our success as a species. In this Perspective, we explore the notion of 'machine culture', culture mediated or generated by machines. We argue that intelligent machines simultaneously transform the cultural evolutionary processes of variation, transmission and selection. Recommender algorithms are altering social learning dynamics. Chatbots are forming a new mode of cultural transmission, serving as cultural models. Furthermore, intelligent machines are evolving as contributors in generating cultural traits-from game strategies and visual art to scientific results. We provide a conceptual framework for studying the present and anticipated future impact of machines on cultural evolution, and present a research agenda for the study of machine culture., (© 2023. Springer Nature Limited.)

Puritanism may evolve into a technological variant based on norms of delegation of actions and perceptions to artificial intelligence. Instead of training self-control, people may be expected to cede their agency to self-controlled machines. The cost-benefit balance of this machine puritanism may be less aversive to wealthy individualistic democracies than the old puritanism they have abandoned.

With the progress of artificial intelligence and the emergence of global online communities, humans and machines are increasingly participating in mixed collectives in which they can help or hinder each other. Human societies have had thousands of years to consolidate the social norms that promote cooperation; but mixed collectives often struggle to articulate the norms which hold when humans coexist with machines. In five studies involving 7917 individuals, we document the way people treat machines differently than humans in a stylized society of beneficiaries, helpers, punishers, and trustors. We show that a different amount of trust is gained by helpers and punishers when they follow norms over not doing so. We also demonstrate that the trust-gain of norm-followers is associated with trustors' assessment about the consensual nature of cooperative norms over helping and punishing. Lastly, we establish that, under certain conditions, informing trustors about the norm-consensus over helping tends to decrease the differential treatment of both machines and people interacting with them. These results allow us to anticipate how humans may develop cooperative norms for human-machine collectives, specifically, by relying on already extant norms in human-only groups. We also demonstrate that this evolution may be accelerated by making people aware of their emerging consensus., (© 2023. The Author(s).)

Artificial intelligence (AI) can be harnessed to create sophisticated social and moral scoring systems-enabling people and organizations to form judgments of others at scale. However, it also poses significant ethical challenges and is, subsequently, the subject of wide debate. As these technologies are developed and governing bodies face regulatory decisions, it is crucial that we understand the attraction or resistance that people have for AI moral scoring. Across four experiments, we show that the acceptability of moral scoring by AI is related to expectations about the quality of those scores, but that expectations about quality are compromised by people's tendency to see themselves as morally peculiar. We demonstrate that people overestimate the peculiarity of their moral profile, believe that AI will neglect this peculiarity, and resist for this reason the introduction of moral scoring by AI., (© The Author(s) 2023. Published by Oxford University Press on behalf of National Academy of Sciences.)

Objective. When medical resources are scarce, clinicians must make difficult triage decisions. When these decisions affect public trust and morale, as was the case during the COVID-19 pandemic, experts will benefit from knowing which triage metrics have citizen support. Design. We conducted an online survey in 20 countries, comparing support for 5 common metrics (prognosis, age, quality of life, past and future contribution as a health care worker) to a benchmark consisting of support for 2 no-triage mechanisms (first-come-first-served and random allocation). Results. We surveyed nationally representative samples of 1000 citizens in each of Brazil, France, Japan, and the United States and also self-selected samples from 20 countries (total N = 7599) obtained through a citizen science website (the Moral Machine). We computed the support for each metric by comparing its usability to the usability of the 2 no-triage mechanisms. We further analyzed the polarizing nature of each metric by considering its usability among participants who had a preference for no triage. In all countries, preferences were polarized, with the 2 largest groups preferring either no triage or extensive triage using all metrics. Prognosis was the least controversial metric. There was little support for giving priority to healthcare workers. Conclusions. It will be difficult to define triage guidelines that elicit public trust and approval. Given the importance of prognosis in triage protocols, it is reassuring that it is the least controversial metric. Experts will need to prepare strong arguments for other metrics if they wish to preserve public trust and morale during health crises., Highlights: We collected citizen preferences regarding triage decisions about scarce medical resources from 20 countries.We find that citizen preferences are universally polarized.Citizens either prefer no triage (random allocation or first-come-first served) or extensive triage using all common triage metrics, with "prognosis" being the least controversial.Experts will need to prepare strong arguments to preserve or elicit public trust in triage decisions., Competing Interests: The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Jean-François Bonnefon acknowledges support from the grant ANR-17-EURE-0010 and the research foundation TSE-Partnership. The funding agreement ensured the authors’ independence in designing the study, interpreting the data, writing, and publishing the report., (© The Author(s) 2022.)

How does the public want a COVID-19 vaccine to be allocated? We conducted a conjoint experiment asking 15,536 adults in 13 countries to evaluate 248,576 profiles of potential vaccine recipients who varied randomly on five attributes. Our sample includes diverse countries from all continents. The results suggest that in addition to giving priority to health workers and to those at high risk, the public favors giving priority to a broad range of key workers and to those with lower income. These preferences are similar across respondents of different education levels, incomes, and political ideologies, as well as across most surveyed countries. The public favored COVID-19 vaccines being allocated solely via government programs but were highly polarized in some developed countries on whether taking a vaccine should be mandatory. There is a consensus among the public on many aspects of COVID-19 vaccination, which needs to be taken into account when developing and communicating rollout strategies., Competing Interests: The authors declare no competing interest., (Copyright © 2021 the Author(s). Published by PNAS.)

Human interactions often involve a choice between acting selfishly (in ones' own interest) and acting prosocially (in the interest of others). Fast and slow models of prosociality posit that people intuitively favor 1 of these choices (the selfish choice in some models, the prosocial choice in other models) and need to correct this intuition through deliberation to make the other choice. We present 7 studies that force us to reconsider this longstanding corrective dual-process view. Participants played various economic games in which they had to choose between a prosocial and a selfish option. We used a 2-response paradigm in which participants had to give their first, initial response under time pressure and cognitive load. Next, participants could take all the time they wanted to reflect on the problem and give a final response. This allowed us to identify the intuitively generated response that preceded the final response given after deliberation. Results consistently showed that both prosocial and selfish responses were predominantly made intuitively rather than after deliberate correction. Pace the deliberate correction view, the findings indicate that making prosocial and selfish choices does typically not rely on different types of reasoning modes (intuition vs. deliberation) but rather on different types of intuitions. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

As machines powered by artificial intelligence (AI) influence humans' behaviour in ways that are both like and unlike the ways humans influence each other, worry emerges about the corrupting power of AI agents. To estimate the empirical validity of these fears, we review the available evidence from behavioural science, human-computer interaction and AI research. We propose four main social roles through which both humans and machines can influence ethical behaviour. These are: role model, advisor, partner and delegate. When AI agents become influencers (role models or advisors), their corrupting power may not exceed the corrupting power of humans (yet). However, AI agents acting as enablers of unethical behaviour (partners or delegates) have many characteristics that may let people reap unethical benefits while feeling good about themselves, a potentially perilous interaction. On the basis of these insights, we outline a research agenda to gain behavioural insights for better AI oversight.

Machines do not 'think fast and slow' in the sense that humans do in dual-process models of cognition. However, the people who create the machines may attempt to emulate or simulate these fast and slow modes of thinking, which will in turn affect the way end users relate to these machines. In this opinion article we consider the complex interplay in the way various stakeholders (engineers, user experience designers, regulators, ethicists, and end users) can be inspired, challenged, or misled by the analogy between the fast and slow thinking of humans and the Fast and Slow Thinking of machines., (Copyright © 2020 Elsevier Ltd. All rights reserved.)

Competing Interests: The authors declare no competing interest.

When do people find it acceptable to sacrifice one life to save many? Cross-cultural studies suggested a complex pattern of universals and variations in the way people approach this question, but data were often based on small samples from a small number of countries outside of the Western world. Here we analyze responses to three sacrificial dilemmas by 70,000 participants in 10 languages and 42 countries. In every country, the three dilemmas displayed the same qualitative ordering of sacrifice acceptability, suggesting that this ordering is best explained by basic cognitive processes rather than cultural norms. The quantitative acceptability of each sacrifice, however, showed substantial country-level variations. We show that low relational mobility (where people are more cautious about not alienating their current social partners) is strongly associated with the rejection of sacrifices for the greater good (especially for Eastern countries), which may be explained by the signaling value of this rejection. We make our dataset fully available as a public resource for researchers studying universals and variations in human morality., Competing Interests: The authors declare no competing interest., (Copyright © 2020 the Author(s). Published by PNAS.)

When an automated car harms someone, who is blamed by those who hear about it? Here we asked human participants to consider hypothetical cases in which a pedestrian was killed by a car operated under shared control of a primary and a secondary driver and to indicate how blame should be allocated. We find that when only one driver makes an error, that driver is blamed more regardless of whether that driver is a machine or a human. However, when both drivers make errors in cases of human-machine shared-control vehicles, the blame attributed to the machine is reduced. This finding portends a public under-reaction to the malfunctioning artificial intelligence components of automated cars and therefore has a direct policy implication: allowing the de facto standards for shared-control vehicles to be established in courts by the jury system could fail to properly regulate the safety of those vehicles; instead, a top-down scheme (through federal laws) may be called for.

Bows and arrows, houses and kayaks are just a few examples of the highly optimized tools that humans have produced and used to colonize new environments 1,2 . Because there is much evidence that humans' cognitive abilities are unparalleled 3,4 , many believe that such technologies resulted from our superior causal reasoning abilities 5-7 . However, others have stressed that the high dimensionality of human technologies makes them very difficult to understand causally 8 . Instead, they argue that optimized technologies emerge through the retention of small improvements across generations without requiring understanding of how these technologies work 1,9 . Here we show that a physical artefact becomes progressively optimized across generations of social learners in the absence of explicit causal understanding. Moreover, we find that the transmission of causal models across generations has no noticeable effect on the pace of cultural evolution. The reason is that participants do not spontaneously create multidimensional causal theories but, instead, mainly produce simplistic models related to a salient dimension. Finally, we show that the transmission of these inaccurate theories constrains learners' exploration and has downstream effects on their understanding. These results indicate that complex technologies need not result from enhanced causal reasoning but, instead, can emerge from the accumulation of improvements made across generations.

Machines powered by artificial intelligence increasingly mediate our social, cultural, economic and political interactions. Understanding the behaviour of artificial intelligence systems is essential to our ability to control their actions, reap their benefits and minimize their harms. Here we argue that this necessitates a broad scientific research agenda to study machine behaviour that incorporates and expands upon the discipline of computer science and includes insights from across the sciences. We first outline a set of questions that are fundamental to this emerging field and then explore the technical, legal and institutional constraints on the study of machine behaviour.

According to the "1-in-X" effect, "1-in-X" ratios (e.g., 1 in 12) trigger a higher subjective probability than do numerically equivalent "N-in-X*N" ratios (e.g., 3 in 36). Here we tested the following: (a) the effect on objective measures, (b) its consequences for decision-making, (c) whether this effect is a form of bias by measuring probability accuracy, and (d) its amplification in people with lower health literacy and numeracy. In parallel-designed experiments, 975 participants from the general adult population participated in 1 of 5 experiments following a 2(format: "1-in-X" or "N-in-X*N") × 4(scenarios) mixed design. Participants assessed the risk of contracting a disease on either a verbal probability scale (Experiment 1) or a numerical probability/frequency scale with immediate (Experiments 2-3) or delayed presentation (Experiments 4-5). Participants also made a health-related decision and completed a health literacy and numeracy scale. The "1-in-X" ratios yielded higher probability perceptions than did the "N-in-X*N" ratios and affected relevant decisions. Critically, the "1-in-X" ratios led to a larger objective overestimation of numerical probabilities than did the "N-in-X*N" ratios. People with lower levels of health literacy and numeracy were not more sensitive to the bias. Health professionals should use "1-in-X" ratios with great caution when communicating to patients, because they overestimate health risks. (PsycINFO Database Record (c) 2018 APA, all rights reserved).

With the rapid development of artificial intelligence have come concerns about how machines will make moral decisions, and the major challenge of quantifying societal expectations about the ethical principles that should guide machine behaviour. To address this challenge, we deployed the Moral Machine, an online experimental platform designed to explore the moral dilemmas faced by autonomous vehicles. This platform gathered 40 million decisions in ten languages from millions of people in 233 countries and territories. Here we describe the results of this experiment. First, we summarize global moral preferences. Second, we document individual variations in preferences, based on respondents' demographics. Third, we report cross-cultural ethical variation, and uncover three major clusters of countries. Fourth, we show that these differences correlate with modern institutions and deep cultural traits. We discuss how these preferences can contribute to developing global, socially acceptable principles for machine ethics. All data used in this article are publicly available.

Since Alan Turing envisioned artificial intelligence, technical progress has often been measured by the ability to defeat humans in zero-sum encounters (e.g., Chess, Poker, or Go). Less attention has been given to scenarios in which human-machine cooperation is beneficial but non-trivial, such as scenarios in which human and machine preferences are neither fully aligned nor fully in conflict. Cooperation does not require sheer computational power, but instead is facilitated by intuition, cultural norms, emotions, signals, and pre-evolved dispositions. Here, we develop an algorithm that combines a state-of-the-art reinforcement-learning algorithm with mechanisms for signaling. We show that this algorithm can cooperate with people and other algorithms at levels that rival human cooperation in a variety of two-player repeated stochastic games. These results indicate that general human-machine cooperation is achievable using a non-trivial, but ultimately simple, set of algorithmic mechanisms.

Economic interactions often imply to gauge the trustworthiness of others. Recent studies showed that when making trust decisions in economic games, people have some accuracy in detecting trustworthiness from the facial features of unknown partners. Here we provide evidence that this face-based trustworthiness detection is a fast and intuitive process by testing its performance at split-second levels of exposure. Participants played a Trust game, in which they made decisions whether to trust another player based on their picture. In two studies, we manipulated the exposure time of the picture. We observed that trustworthiness detection remained better than chance for exposure times as short as 100 ms, although it disappeared with an exposure time of 33 ms. We discuss implications for ongoing debates on the use of facial inferences for social and economic decisions.

Research on trustworthiness perception from faces has unfolded in a way that is strikingly reminiscent of Jussim's narrative in his 2012 book. Jussim's analysis warns us against overemphasizing evidence about prejudice over evidence about accuracy, when both are scant; and reminds us to hold all accounts to the same standards, whether they call on societal biases or true signals.

Autonomous vehicles (AVs) should reduce traffic accidents, but they will sometimes have to choose between two evils, such as running over pedestrians or sacrificing themselves and their passenger to save the pedestrians. Defining the algorithms that will help AVs make these moral decisions is a formidable challenge. We found that participants in six Amazon Mechanical Turk studies approved of utilitarian AVs (that is, AVs that sacrifice their passengers for the greater good) and would like others to buy them, but they would themselves prefer to ride in AVs that protect their passengers at all costs. The study participants disapprove of enforcing utilitarian regulations for AVs and would be less willing to buy such an AV. Accordingly, regulating for utilitarian algorithms may paradoxically increase casualties by postponing the adoption of a safer technology., (Copyright © 2016, American Association for the Advancement of Science.)

We identify a blind spot in the early Theory of Mind processing of conditional sentences that describe a protagonist's potential action, and its predictable consequences. We propose that such sentences create expectations through two independent channels. A decision theoretic channel creates an expectation that the action will be taken (viz., not taken) if it has desirable (viz., undesirable) consequences, but a structural channel acts in parallel to create an expectation that the action will be taken, irrespective of desirability. Accordingly, reading should be disrupted when a protagonist avoids an action with desirable consequences, but reading should not be disrupted when a protagonist takes an action with undesirable consequences. This prediction was supported by the eye movements of participants reading systematically varied vignettes. Reading was always disrupted when the protagonist avoided an action with desirable consequences, but disruptions were either delayed (Experiment 1) or recovered from faster (Experiment 2) when the protagonist took an action with undesirable consequences., (Copyright © 2015 Elsevier B.V. All rights reserved.)

For unknown reasons, individuals who are confident in their intuitions are more likely to hold supernatural beliefs. How does an intuitive cognitive style lead one to believe in faith healing, astrology, or extrasensory perception (ESP)? We hypothesize that cognitive style is critically important after one experiences an uncanny event that seems to invite a supernatural explanation. In three studies, we show that irrespective of their prior beliefs in the supernatural, non-reflective thinkers are more likely than reflective thinkers to accept supernatural causation after an uncanny encounter with astrology and ESP. This is the first time that controlled experiments demonstrate the negative dynamics of reflection and supernatural causality attribution. We consider the possible generalization of our findings to religious beliefs and their implications for the social vulnerability of non-reflective individuals., (© 2015 by the Society for Personality and Social Psychology, Inc.)

Remembering the past and imagining the future both rely on complex mental imagery. We considered the possibility that constructing a future scene might tap a component of mental imagery that is not as critical for remembering past scenes. Whereas visual imagery plays an important role in remembering the past, we predicted that spatial imagery plays a crucial role in imagining the future. For the purpose of teasing apart the different components underpinning scene construction in the two experiences of recalling episodic memories and shaping novel future events, we used a paradigm that might selectively affect one of these components (i.e., the spatial). Participants performed concurrent eye movements while remembering the past and imagining the future. These concurrent eye movements selectively interfere with spatial imagery, while sparing visual imagery. Eye movements prevented participants from imagining complex and detailed future scenes, but had no comparable effect on the recollection of past scenes. Similarities between remembering the past and imagining the future are coupled with some differences. The present findings uncover another fundamental divergence between the two processes.

We examine how the beliefs and desires of a protagonist are used by readers to predict their intentions as a narrative vignette unfolds. Eye movement measures revealed that readers rapidly inferred an intention when the protagonist desired an outcome, even when this inference was not licensed by the protagonist's belief state. Reading was immediately disrupted when participants encountered a described action that contradicted this inference. During intermediate processing, desire inferences were moderated by the protagonist's belief state. Effects that emerged later in the text were again driven solely by the protagonist's desires. These data suggest that desire-based inferences are initially drawn irrespective of belief state, but are then quickly inhibited if not licensed by relevant beliefs. This inhibition of desire-based inferences may be an effortful process as it was not systematically sustained in later steps of processing.

It has long been known that eye movements are functionally involved in the generation and maintenance of mental images. Indeed, a number of studies demonstrated that voluntary eye movements interfere with mental imagery tasks (e.g., Laeng and Teodorescu in Cogn Sci 26:207-231, 2002). However, mental imagery is conceived as a multifarious cognitive function with at least two components, a spatial component and a visual component. The present study investigated the question of whether eye movements disrupt mental imagery in general or only its spatial component. We present data on healthy young adults, who performed visual and spatial imagery tasks concurrently with a smooth pursuit. In line with previous literature, results revealed that eye movements had a strong disruptive effect on spatial imagery. Moreover, we crucially demonstrated that eye movements had no disruptive effect when participants visualized the depictive aspects of an object. Therefore, we suggest that eye movements serve to a greater extent the spatial than the visual component of mental imagery.

Current computational models of theory of mind typically assume that humans believe each other to selfishly maximize utility, for a conception of utility that makes it indistinguishable from personal gains. We argue that this conception is at odds with established facts about human altruism, as well as the altruism that humans expect from each other. We report two experiments showing that people expect other agents to selfishly maximize their pleasure, even when these other agents behave altruistically. Accordingly, defining utility as pleasure permits us to reconcile the assumption that humans expect each other to selfishly maximize utility with the fact that humans expect each other to behave altruistically.