Objective Trustworthiness Research Articles

First impressions of a financial AI assistant: differences between high trust and low trust users.

Calibrating appropriate trust of non-expert users in artificial intelligence (AI) systems is a challenging yet crucial task. To align subjective levels of trust with the objective trustworthiness of a system, users need information about its strengths and weaknesses. The specific explanations that help individuals avoid over- or under-trust may vary depending on their initial perceptions of the system. In an online study, 127 participants watched a video of a financial AI assistant with varying degrees of decision agency. They generated 358 spontaneous text descriptions of the system and completed standard questionnaires from the Trust in Automation and Technology Acceptance literature (including perceived system competence, understandability, human-likeness, uncanniness, intention of developers, intention to use, and trust). Comparisons between a high trust and a low trust user group revealed significant differences in both open-ended and closed-ended answers. While high trust users characterized the AI assistant as more useful, competent, understandable, and humanlike, low trust users highlighted the system's uncanniness and potential dangers. Manipulating the AI assistant's agency had no influence on trust or intention to use. These findings are relevant for effective communication about AI and trust calibration of users who differ in their initial levels of trust.

Self-equity as a trustworthiness measure: The relationship among self-equity and security clearance eligibility adjudications in US Army recruits

ABSTRACT The United States government’s clearance adjudication process examines past behavior to determine soldier eligibility for a security clearance. For young recruits with a short-documented history, however, little information is available. While informal social controls generally associate with criminal desistance, desistance speaks little about those who have yet to offend. This work extends informal social control theory to better understand military clearance eligibility adjudication outcomes as measured in terms of a self-equity construct. This analysis looks at a twelve-year cohort of US Army recruits who received clearance eligibility adjudication within the first five years of service, as recorded in military archival data in the Person-event Data Environment (PDE) database. Laub, Rowan, & Sampson’s (2018) age-graded theory of informal social control is tested to estimate models, capturing the self-equity effects of moral waiver, qualification-test percentiles, service time, rank, education, and childbirth–marriage interaction. The results demonstrate that self-equity substantively relates to security clearance eligibility adjudication outcomes. These findings carry policy implications for the creation of an objective trustworthiness measure in the absence of reliable documented history.

Can processing of face trustworthiness bypass early visual cortex? A transcranial magnetic stimulation masking study

As a highly social species, we constantly evaluate human faces to decide whether we can trust someone. Previous studies suggest that face trustworthiness can be processed unconsciously, but the underlying neural pathways remain unclear. Specifically, the question remains whether processing of face trustworthiness relies on early visual cortex (EVC), required for conscious perception. If processing of trustworthiness can bypass EVC, then disrupting EVC should impair subjective (conscious) trustworthiness perception while leaving objective (forced-choice) trustworthiness judgment intact. We applied double-pulse transcranial magnetic stimulation (TMS) to right EVC, at different stimulus onset asynchronies (SOAs) from presentation of a face in either the left or right hemifield. Faces were slightly rotated clockwise or counterclockwise, and were either trustworthy or untrustworthy. On each trial, participants discriminated 1) trustworthiness, 2) stimulus rotation, and 3) reported subjective visibility of trustworthiness. At early SOAs and specifically in the left hemifield, performance on the rotation task was impaired by TMS. Crucially, though TMS also impaired subjective visibility of trustworthiness, no effects on trustworthiness discrimination were obtained. Thus, conscious perception of face trustworthiness (captured by subjective visibility ratings) relies on intact EVC, while objective forced-choice trustworthiness judgments may not. These results are consistent with the hypothesis that objective trustworthiness processing can bypass EVC. For basic visual features, extrastriate pathways are well-established; but face trustworthiness depends on a complex configuration of features. Its potential processing without EVC is therefore of particular interest, further highlighting its ecological relevance.

An approach for public cloud trustworthiness assessment based on users' evaluation and performance indicators

Aiming at the problem of how to quantify the trustworthiness of public cloud services, this paper proposes a method to evaluate the trustworthiness of public clouds based on users' subjective evaluations and objective monitoring values of performance indicators. According to the objective values obtained from monitoring organisations, the objective trustworthiness is calculated by using a multi-attribute decision-making method based on the ideal point (TOPSIS). Meanwhile, qualitative users evaluations are quantified by exploiting the cloud model. Finally, a comprehensive assessment of the trustworthiness of cloud service is formed based on the above results. To meet the personalised requirements, users could configure the weight of parameters included in the proposed algorithms, ranking cloud services for specific needs. Compared with other assessment algorithms, both the subjective source and objective source are taken into account, the customised weight scheme for users is provided, the model and the algorithms are designed, and a prototype is developed.

An approach for public cloud trustworthiness assessment based on users' evaluation and performance indicators

Aiming at the problem of how to quantify the trustworthiness of public cloud services, this paper proposes a method to evaluate the trustworthiness of public clouds based on users' subjective evaluations and objective monitoring values of performance indicators. According to the objective values obtained from monitoring organisations, the objective trustworthiness is calculated by using a multi-attribute decision-making method based on the ideal point (TOPSIS). Meanwhile, qualitative users evaluations are quantified by exploiting the cloud model. Finally, a comprehensive assessment of the trustworthiness of cloud service is formed based on the above results. To meet the personalised requirements, users could configure the weight of parameters included in the proposed algorithms, ranking cloud services for specific needs. Compared with other assessment algorithms, both the subjective source and objective source are taken into account, the customised weight scheme for users is provided, the model and the algorithms are designed, and a prototype is developed.

A Framework for Systematic Refinement of Trustworthiness Requirements

The trustworthiness of systems that support complex collaborative business processes is an emergent property. In order to address users’ trust concerns, trustworthiness requirements of software systems must be elicited and satisfied. The aim of this paper is to address the gap that exists between end-users’ trust concerns and the lack of implementation of proper trustworthiness requirements. New technologies like cloud computing bring new capabilities for hosting and offering complex collaborative business operations. However, these advances might bring undesirable side effects, e.g., introducing new vulnerabilities and threats caused by collaboration and data exchange over the Internet. Hence, users become more concerned about trust. Trust is subjective; trustworthiness requirements for addressing trust concerns are difficult to elicit, especially if there are different parties involved in the business process. We propose a user-centered trustworthiness requirement analysis and modeling framework. We integrate the subjective trust concerns into goal models and embed them into business process models as objective trustworthiness requirements. Business process model and notation is extended to enable modeling trustworthiness requirements. This paper focuses on the challenges of elicitation, refinement and modeling trustworthiness requirements. An application example from the healthcare domain is used to demonstrate our approach.

Trust in Technology as a Safety Aspect in Highly Automated Driving

Abstract Trust in technology is an important factor to be considered for safety-critical systems. Of particular interest today is the transport domain, as more and more complex information and assistance systems find their way into vehicles. Research in driving automation / automated driving systems is in the focus of many research institutes worldwide. On the operational side, active safety systems employed to save lives are frequently used by non-professional drivers that neither know system boundaries nor the underlying functional principle. This is a serious safety issue, as systems are activated under false circumstances and with wrong expectations. At least some of the recent incidents with advanced driving assistance systems (ADAS) or automated driving systems (ADS; SAE J3016) could have been prevented with a full understanding of the driver about system functionality and limitations (instead of overreliance). Drivers have to be trained to accept and use these systems in a way, that subjective trust matches objective trustworthiness (cf. “appropriate trust”) to prevent disuse and / or misuse. In this article, we present an interaction model for trust calibration that issues personalized messages in real time. On the showcase of automated driving we report the results of two user studies related to trust in ADS and driving ethics. In the first experiment (N = 48), mental and emotional states of front-seat passengers were compared to get deeper insight into the dispositional trust of potential users of automated vehicles. Using quantitative and qualitative methods, we found that subjects accept and trust ADSs almost similarly as male / female drivers. In another study (N = 40), moral decisions of drivers were investigated in a systematic way. Our results indicate that the willingness of drivers to risk even severe accidents increases with the number and age of pedestrians that would otherwise be sacrificed. Based on our initial findings, we further discuss related aspects of trust in driving automation. Effective shared vehicle control and expected advantages of fully / highly automated driving (SAE levels 3 or higher) can only be achieved when trust issues are demonstrated and resolved.

Design and evaluation of algorithms for obtaining objective trustworthiness on acquaintances in P2P overlay networks

In fully distributed P2P overlay networks, each peer has to obtain information on objects distributed through communicating with its acquaintances. An acquaintance might hold obsolete information due to the propagation delay and faults of peers. Hence, a peer has to collect information only from trustworthy acquaintances. The subjective trustworthiness shows how much a peer trusts its acquaintance. The objective trustworthiness indicates how much other peers trust the acquaintance. We discuss three algorithms to calculate the objective trustworthiness. The algorithm OT1 shows a type of flooding algorithm. A peer sends request messages only to the acquaintances to reduce the number of messages in the algorithms OT2 and OT3. Furthermore, the peer p s sends messages to only acquaintances which the peer p s can trust in OT3. Then, the peer p s receives the subjective trustworthiness on the acquaintance from the acquaintance p 2. The peer p u obtains the average value of the subjective trustworthiness collected as the objective trustworthiness. We evaluate the three algorithms in terms of how correct objective trustworthiness can be obtained by the algorithms.

Trustworthy acquaintances in Peer-to-Peer (P2P) overlay networks

In fully distributed Peer-to-Peer (P2P) overlay networks, each peer has to obtain information on objects distributed through communicating with its acquaintance peers. An acquaintance might hold obsolete information owing to the propagation delay and faults of peers. Hence, a peer has to collect correct object information only from trustworthy acquaintances. The subjective trustworthiness of an acquaintance is obtained by directly interacting with the acquaintance. The objective trustworthiness of an acquaintance is obtained by collecting the subjective trustworthiness from other peers. If a peer is confident of its own subjective trustworthiness, the peer takes the subjective trustworthiness. Otherwise, the peer takes the objective one. We evaluate how the subjective trustworthiness changes in change of satisfiablity of an acquaintance.

From Subjective Trust to Objective Trustworthiness in On-line Social Networks: Overview and Challenges

Nowadays dozens of people share their content in the current Web 2.0 space, talk with friends in social networking sites such as Facebook and live on the Net in many other ways. They do all this quite naturally, forgetting the healthy cautiousness sometimes. In real life we rely on trusted people. Do we know how to reflect real-world trust mechanisms into on-line social software? In the article we focused to bring overview on state of the art in main ideas behind a trust processing in online social networking systems. What are common sources of subjective trust, how the trust emerges and what are the sources of trust dynamics? How can be trust captured into the systems, how can be explicit trust processed to infer indirect trust, the trust between users who do not know each other? And what are the ways to infer objective metrics of trust, the reputation or trustworthiness? Finally, we point out selected challenges related to the trust in current highly dynamic social networks.