Misplaced Trust Research Articles

Users do not trust recommendations from a large language model more than AI-sourced snippets

BackgroundThe ability of large language models to generate general purpose natural language represents a significant step forward in creating systems able to augment a range of human endeavors. However, concerns have been raised about the potential for misplaced trust in the potentially hallucinatory outputs of these models.ObjectivesThe study reported in this paper is a preliminary exploration of whether trust in the content of output generated by an LLM may be inflated in relation to other forms of ecologically valid, AI-sourced information.MethodParticipants were presented with a series of general knowledge questions and a recommended answer from an AI-assistant that had either been generated by an ChatGPT-3 or sourced by Google’s AI-powered featured snippets function. We also systematically varied whether the AI-assistant’s advice was accurate or inaccurate.ResultsTrust and reliance in LLM-generated recommendations were not significantly higher than that of recommendations from a non-LLM source. While accuracy of the recommendations resulted in a significant reduction in trust, this did not differ significantly by AI-application.ConclusionUsing three predefined general knowledge tasks and fixed recommendation sets from the AI-assistant, we did not find evidence that trust in LLM-generated output is artificially inflated, or that people are more likely to miscalibrate their trust in this novel technology than another commonly drawn on form of AI-sourced information.

Perkins Operations: Tactics Used in Undercover Interactions

In the United States, it is permissible to place an undercover police officer in the jail cell with a suspect. This tactical move is rare and launched only for serious crimes, and it takes place before any charges have been filed. This tactic goes under the name of Perkins operations, from the case ruling that if an individual speaks freely to someone whom they believe is a fellow inmate it is allowed to take advantage of their misplaced trust (Illinois v. Perkins, 1990). In this study, we examine 22 Perkins operations, 60 hours of secretly taped interactions in the cells, and we describe and categorize the different approaches and tactics that the undercover officers used. Based on the descriptive analysis, we conceptualize two pathways to information elicitation (direct and relational) and explore the undercover officers’ use of risky interview tactics. The findings suggest that undercover officers use four broader approaches to establish relationships and gather information, and we were able to identify only a few instances of risky tactics in this sample. The relevance of the findings for human intelligence gathering and counterintelligence are discussed.

Why talk about sin? Luther's understanding of sin and hopeful sin‐talk in the 21st century

AbstractFor Luther talking about sin and the sinful nature of human beings has a strong pastoral significance. His emphasis on the “bondage” of the human will is tightly connected to his insistence on the human sinful condition, and our inability to choose to be or not to be held captive by sin. My conclusion is that it is indeed important to continue to talk about sin if the Christian discourse about God's forgiveness and grace is to make sense. Furthermore, I believe Luther's understanding of sin as misplaced trust, the distinction he makes between sin and sins, and his idea of a justified sinner can indeed make a significant contribution to a hopeful sin‐talk within Christian communities today. It is, however, necessary to pay attention to Luther's historical context and to reevaluate his understanding of human nature and human sinfulness from a feminist critical perspective.The sin‐concept has gradually been losing its relevance within Christian communities. Therefore the question: why should we continue to talk about sin? The aim of this article is to explore Luther's understanding of sin and human sinfulness, in order to find out if, and then how, he might prove helpful when it comes to the interpretation of the concept of sin in the 21st century. The focus is on Luther's pastoral writings in The Small and The Large Catechism (1529), together with his Smalcald Articles (1537).

A review of experimental research on organizational trust

ABSTRACT Trust profoundly shapes organisational, group, and dyadic outcomes. Reflecting its importance, a substantial and growing body of scholarship has investigated the topic of trust. Much of this work has used experiments to identify clear, causal relationships. However, in contrast to theoretical work that conceptualises trust as a multi-faceted (e.g. ability, benevolence, integrity), multi-level (e.g. interpersonal, intergroup), and dynamic construct, experimental scholarship investigating trust has largely investigated benevolence-based trust in dyadic relationships. As a result of the relatively limited set of paradigms experimental scholars have used to investigate trust, many questions related to different forms and types of trust remain un- and under-explored in experimental work. In this review, we take stock of the existing experimental trust scholarship and identify key gaps in our current understanding of trust. We call for future experimental work to investigate ability-based and integrity-based trust, to advance our understanding of the interplay between relationship history and trust, to study trust as a multi-level construct, to focus on the consequences of trust including the hazards of misplaced trust, and to study trust maintenance. To support these lines of inquiry, we introduce an ideal-typical process model to develop or adapt appropriate trust experiments.

The views about Big Data among professionals of police forces: A scoping review of empirical studies

There is a growing tendency toward the introduction of Big Data in police departments to improve the prevention and investigation of crime. However, there is little systematic knowledge about the perspectives of professionals in police forces regarding this technology. This article fills this gap by presenting a scoping review that systematizes empirical studies of the views of professionals in police forces about Big Data in the field of policing and criminal investigation. Fourteen articles were analyzed following a descriptive–analytical method. Optimistic and oppositional views about Big Data among professionals in police forces were then described. Optimistic views focused on the potential of Big Data to improve the objectivity and efficiency of policing and better manage police resources by providing new capabilities and strategies by which to perform crime predictions, risk assessments, criminal investigations, crime analysis, risk management, cooperation and data exchange. Oppositional views related to the police's awareness of Big Data's biases, the lack of a regulatory landscape, misuse of data, privacy threats, data security, misplaced trust in technology, the absence of major changes in work practices, and practical barriers. Analysis of the co-presence of optimistic and oppositional views adds a comprehensive and interpretative argument that might contribute to critical reflections on the technological feasibility, societal usability, and desirability of Big Data technologies in police departments.

FINANCIAL FRAUD EXPERIENCES AT AGE 85+: AN EXPLORATORY STUDY WITH THE MIT AGELAB 85+ LIFESTYLE LEADERS PANEL

Abstract Financial exploitation of older adults has emerged as one of the most common and costly forms of elder abuse. While prior research has examined the risk and protective factors that contribute to older adults’ experiences of being defrauded, less has specifically explored the prevalence of these factors among the 85+ age demographic. This not only includes hearing from the 85+ directly about the factors that have contributed to successful financial exploitation, but also the factors that contributed to their experiences of unsuccessful and ‘near miss’ exploitation. This paper shares findings from a mixed methods study with the MIT AgeLab 85+ Lifestyle Leaders, a U.S.-based panel of octogenarians and nonagenarians. Leveraging an online survey (n=32) and virtual focus groups (n=18) conducted in September 2021, findings distinguish among three categories of experiences with financial fraud – those had been, almost were, or have never been a victim of financial fraud within the past five years. In particular, ‘almost victims’ frequently scored in between ‘have been’ and ‘have never been’ victims with regard to several questionnaire variables. Six perceived risk factors for financial fraud victimization emerged from both data sources, including being older, cognitive decline, misplaced trust and a lack of knowledge, loneliness, technology use, and stereotypes of older adults. The majority of Lifestyle Leaders identified protective factors including, knowledge about protecting oneself from financial scams, a lifetime of experience and one’s social network. These findings will be discussed in further detail along with their implications for multidisciplinary professionals working with older adults.

Costly Displays in a Digital World: Signalling Trustworthiness on Social Media

ABSTRACT Placing our trust wisely is both difficult and important. The challenge of knowing who to trust inheres at least partially in the fact that coinciding interests cannot be taken for granted, and that language, as the principal medium through which would-be interactants make their interests known, doesn’t discriminate between true and feigned proclamations of good intent. Because our patterns of trust partition the world into reliable and unreliable sources, trust is also important: it determines how we distribute our social and epistemic dependencies, which in turn informs much of what we believe and do. The fact that our interactions with the world and each other are increasingly mediated by digital communication technologies affects both these elements of trust: though unabated access to the internet invites us to distribute our dependencies more broadly, the kinds of information environments that we encounter online make misplaced trust a more common occurrence. I argue that a large part of the problem, as well as one possible set of solutions, resides in a deceptively simple dictum: the internet makes talk cheap, and cheap talk can’t be trusted.

Misplaced Trust and Distrust: How Not to Engage with Medical Artificial Intelligence.

Artificial intelligence (AI) plays a rapidly increasing role in clinical care. Many of these systems, for instance, deep learning-based applications using multilayered Artificial Neural Nets, exhibit epistemic opacity in the sense that they preclude comprehensive human understanding. In consequence, voices from industry, policymakers, and research have suggested trust as an attitude for engaging with clinical AI systems. Yet, in the philosophical and ethical literature on medical AI, the notion of trust remains fiercely debated. Trust skeptics hold that talking about trust in nonhuman agents constitutes a category error and worry about the concept being misused for ethics washing. Proponents of trust have responded to these worries from various angles, disentangling different concepts and aspects of trust in AI, potentially organized in layers or dimensions. Given the substantial disagreements across these accounts of trust and the important worries about ethics washing, we embrace a diverging strategy here. Instead of aiming for a positive definition of the elements and nature of trust in AI, we proceed ex negativo, that is we look at cases where trust or distrust are misplaced. Comparing these instances with trust expedited in doctor-patient relationships, we systematize these instances and propose a taxonomy of both misplaced trust and distrust. By inverting the perspective and focusing on negative examples, we develop an account that provides useful ethical constraints for decisions in clinical as well as regulatory contexts and that highlights how we should not engage with medical AI.

Adolescents’ self-efficacy and digital health literacy: a cross-sectional mixed methods study

BackgroundThe internet and social media are increasingly popular sources of health information for adolescents. Using online health information requires digital health literacy, consisting of literacy, analytical skills and personal capabilities such as self-efficacy. Appraising trustworthiness and relevance of online health information requires critical health literacy to discriminate between sources, critically analyse meaning and relevance, and use information for personal health. Adolescents with poor digital health literacy risk using misinformation, with potential negative health outcomes. We aimed to understand adolescents’ contemporary digital health literacy and compared self-efficacy with capability.MethodsAdolescents (12–17 years) completed an eHEALS self-report digital health literacy measure, a practical search task using a think-aloud protocol and an interview to capture perceived and actual digital health literacy. eHEALS scores were generated using descriptive statistics, search tasks were analysed using an observation checklist and interviews were thematically analysed based on Social Cognitive Theory, focussing on self-efficacy.ResultsTwenty-one participants generally had high self-efficacy using online health information but perceived their digital health literacy to be higher than demonstrated. They accessed online health information unintentionally on social media and intentionally via search engines. They appraised information medium, source and content using general internet searching heuristics taught at school. Information on social media was considered less trustworthy than websites, but participants used similar appraisal strategies for both; some search/appraisal heuristics were insufficiently nuanced for digital health information, sometimes resulting in misplaced trust or diminished self-efficacy. Participants felt anxious or relieved after finding online health information, depending on content, understanding and satisfaction. They did not act on information without parental and/or health professional advice. They rarely discussed findings with health professionals but would welcome discussions and learning how to find and appraise online health information.ConclusionsWhilst adolescents possess many important digital health literacy skills and generally feel self-efficacious in using them, their critical health literacy needs improving. Adolescents desire increased digital health literacy so they can confidently appraise health information they find online and on social media. Co-designed educational interventions with adolescents and health providers are required.

Understanding online grooming: An interpretative phenomenological analysis of adolescents' offline meetings with adult perpetrators

BackgroundIt is the norm for adolescents to form friendships online but connecting with strangers poses risks of online grooming. Adult perpetrators take advantage of the internet's anonymity and accessibility to befriend and groom young people for sexual abuse. Much of the existing research has focused on the offender's perspective and grooming tactics, while extensive studies on victim profiles have focused on risk factors and young people's vulnerabilities to online grooming. ObjectiveThis research takes an in-depth look into grooming processes from the perspective of the adolescent victim and explores how the relationships progressed online-to-offline, resulting in sexual abuse. It aims to understand young people's motivations and mindsets by exploring their lived experiences of offline meeting with adult perpetrators. Participants and settingThe sample consisted of adolescents aged 14–21 in the UK and EU. MethodsUsing data from the Risk-taking Online Behavior Empowerment through Research and Training Project, six interviews were examined using an interpretative phenomenological analysis. ResultsFour themes emerged: 1) Varied interpretations of risk 2) Taking control 3) Seeking security and validation; and 4) Shifts in perspective. The subtleties of online grooming were masked by the adolescents' misplaced trust as they displayed agency in pursuing opportunities for online friendships or sexual activities. ConclusionsFindings suggest an insufficient knowledge of grooming and the need to more clearly define risks among adolescents. Safety guidelines should highlight the subtleties of grooming tactics and harmful behaviours that may not be readily perceived as abuse.

Is anybody in there? Towards a model of affect and trust in human–AI information interactions

Advancements in search engines that utilize machine learning increase the likelihood that users will perceive these systems as worthy of trust.Methods. The nature and implications of trust in the context of algorithmic systems that utilize machine learning is examined and the resulting conception of trust is modelled. While current artificial intelligence does not meet the requirements of moral autonomy necessary to be considered trustworthy, people may still engage in misplaced trust based on the perception of moral autonomy. Users who place their trust in algorithmic systems limit their critical engagement with, and assessment of, the information interaction. A preliminary high-level model of trust’s role in information interactions adapting Ingwersen and Jarvelin’s Integrative Model for Interactive Information Seeking and Retrieval is proposed using the Google search engine as an example. We need to recognize that is it possible for users to react to information systems in a social manner that may lead to the formation of trust attitudes. As information professionals we want to develop interventions that will encourage users to stay critically engaged with their interactions with information systems, even when they perceive them to be autonomous.

Dietary Supplement Use among College Students and Misplaced Trust

Dietary Supplement Use among College Students and Misplaced Trust

Misplaced trust: When trust in science fosters belief in pseudoscience and the benefits of critical evaluation

At a time when pseudoscience threatens the survival of communities, understanding this vulnerability, and how to reduce it, is paramount. Four preregistered experiments (N = 532, N = 472, N = 605, N = 382) with online U.S. samples introduced false claims concerning a (fictional) virus created as a bioweapon, mirroring conspiracy theories about COVID-19, and carcinogenic effects of GMOs (Genetically Modified Organisms). We identify two critical determinants of vulnerability to pseudoscience. First, participants who trust science are more likely to believe and disseminate false claims that contain scientific references than false claims that do not. Second, reminding participants of the value of critical evaluation reduces belief in false claims, whereas reminders of the value of trusting science do not. We conclude that trust in science, although desirable in many ways, makes people vulnerable to pseudoscience. These findings have implications for science broadly and the application of psychological science to curbing misinformation during the COVID-19 pandemic.

Misplaced trust? The relationship between trust, ability to identify commercially influenced results and search engine preference

People have a high level of trust in search engines, especially Google, but only limited knowledge of them, as numerous studies have shown. This leads to the question: To what extent is this trust justified considering the lack of familiarity among users with how search engines work and the business models they are founded on? We assume that trust in Google, search engine preferences and knowledge of result types are interrelated. To examine this assumption, we conducted a representative online survey with n = 2012 German Internet users. We show that users with little search engine knowledge are more likely to trust and use Google than users with more knowledge. A contradiction revealed itself – users strongly trust Google, yet they are unable to adequately evaluate search results. For those users, this may be problematic since it can potentially affect knowledge acquisition. Consequently, there is a need to promote user information literacy to create a more solid foundation for user trust in search engines. The impact of our study lies in emphasising the need for creating appropriate training formats to promote information literacy.

Trust, Trusting and Trustworthiness in Ethical Discourse

Abstract In literary genres concerned with an ethics of character, scholars in the third through fifth/ninth through eleventh century wrote about social trust. In this article I examine the ethical thinking of four such scholars: Ibn Abī l-Dunyā (d. 281/894), al-Kharāʾiṭī (d. 327/939), al-Māwardī (d. 450/1058), and al-Ghazālī (d. 505/1111). Locating conceptualisations of the ethical concerns related to trust and mistrust, trustworthiness, trusting and misplaced trust, in a semantic field including the vocabulary of amāna, tawakkul, thiqa and ḥusn al-ẓann, I identify and discuss in this article some of these concerns, and I analyse in what ways the scholars’ respective conceptualisations of issues related to trust are similar, and how they differ from each other. While some scholars explicitly conceptualise the ethical value of trust in the concept of amāna, the juxtaposed social and spiritual aspects of the concept of amāna is more implicit with other scholars.

Playing with perspective: Narrative voice and trust in Australian independent podcasts

Narrative voice is a frequent site of experimentation in narrative journalism, and in podcasting this voice tends to be more prominent as the listener hears the narrator’s embodied voice. It can build a strong bond with the listener, which is important for independent producers as trust is not automatically afforded to them by association with a trusted media organization. This is particularly true for emerging producers, who also lack a professional reputation. This study examines the techniques used in five Australian independent podcasts to understand how they experiment with narrative voice as a podcast technique to build trust with the listener. It finds trust is built through narrative voice in four specific ways (1) first-person narration, (2) authenticity, (3) empathy and (4) emotional truth. This may allow for a greater variety of voices to be not only heard but trusted within podcasting, but also raises questions about misplaced trust.

We need to talk about deception in social robotics!

Although some authors claim that deception requires intention, we argue that there can be deception in social robotics, whether or not it is intended. By focusing on the deceived rather than the deceiver, we propose that false beliefs can be created in the absence of intention. Supporting evidence is found in both human and animal examples. Instead of assuming that deception is wrong only when carried out to benefit the deceiver, we propose that deception in social robotics is wrong when it leads to harmful impacts on individuals and society. The appearance and behaviour of a robot can lead to an overestimation of its functionality or to an illusion of sentience or cognition that can promote misplaced trust and inappropriate uses such as care and companionship of the vulnerable. We consider the allocation of responsibility for harmful deception. Finally, we make the suggestion that harmful impacts could be prevented by legislation, and by the development of an assessment framework for sensitive robot applications.

Matthew Reidsma. Masked by Trust: Bias in Library Discovery. Sacramento, CA: Library Juice Press, 2019. 194p. Paper, $28.00 (ISBN 978-1-63400-083-3). LCCN 2019010995.

In his book, Masked by Trust: Bias in Library Discovery, Reidsma addresses the increasing use and misplaced trust in search engines, and how library software developers have integrated Google-like elements into discovery systems. Since libraries license these systems, users believe that they are more reliable than web, or general purpose, search engines. However, the author asserts that, since library discovery systems are created by people with biases, they are not as neutral as vendors and libraries claim. Reidsma references several important researchers throughout the text, such as Safiya Umoja Noble, Cathy O’Neil, Carole Cadwalladr, and others, who have published works regarding algorithmic bias. These, in addition to his own research into algorithms in library systems, help support the author’s assertion that library systems are not immune from bias, thereby making the case that even library systems are not neutral.

The Misplaced Trust in the DOJ's Expertise on Criminal Justice Policy

Review of Rachel Elise Barkow's Prisoners of Politics: Breaking the Cycle of Mass Incarceration.

Celebrating Three Decades of Worldwide Stock Market Manipulation

As the decade turns, we reflect on nearly thirty years of successful manipulation of the world's public equity markets. This reflection highlights a few of the key enabling ingredients and lessons learned along the way. A quantitative understanding of market impact and its decay, which we cover briefly, lets you move long-term market prices to your advantage at acceptable cost. Hiding your footprints turns out to be less important than moving prices in the direction most people want them to move. Widespread (if misplaced) trust of market prices - buttressed by overestimates of the cost of manipulation and underestimates of the benefits to certain market participants - makes price manipulation a particularly valuable and profitable tool. Of the many recent stories heralding the dawn of the present golden age of misinformation, the manipulation leading to the remarkable increase in the market capitalization of the world's publicly traded companies over the past three decades is among the best.