Risk and Trust in artificial intelligence technologies: A case study of Autonomous Vehicles

Ajenaghughrure, Ighoyota Ben; Sousa, Sónia; Lamas, David

doi:10.1109/hsi49210.2020.9142686

Cited by 16 publications

(6 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This relatively high accuracy may clarify why several participants were so tolerant toward the system's faulty recommendation, even though it still yielded significantly lower trust ratings. Furthermore, the fact that the interaction posed no risk for the participants may further clarify why some of them were so tolerant, as the literature indicates that trust ratings are the lowest in high-risk situations [16,44,57].…”

Section: Discussionmentioning

confidence: 99%

“…Ref. [16] reports on a study conducted to evaluate the effects of risk perception on trust in automated vehicles. They found that not only did interacting with an automated vehicle in a risky scenario significantly reduce participants' trust and delegation of control to the vehicle, but they also found that initial trust was significantly higher than trust levels after interacting with the vehicle in high-risk conditions.…”

Section: Initial Trustmentioning

confidence: 99%

“…This is relevant with regard to the increasing popularity of models such as neural networks (black box models) and is in light of their intrinsic opaqueness and inscrutability [13,15]. It is all the more relevant in interaction contexts that entail potential risks for the users, such as with automated vehicles [16,17]. In this regard, researchers argue that making the causal chains behind models' decisions interpretable is likely to help people understand the rationales behind those decisions and, importantly, calibrate their expectations and trust [18][19][20].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Trust Development and Explainability: A Longitudinal Study with a Personalized Assistive System

Zafari,

de Pagter,

Papagni

et al. 2024

MTI

View full text Add to dashboard Cite

This article reports on a longitudinal experiment in which the influence of an assistive system’s malfunctioning and transparency on trust was examined over a period of seven days. To this end, we simulated the system’s personalized recommendation features to support participants with the task of learning new texts and taking quizzes. Using a 2 × 2 mixed design, the system’s malfunctioning (correct vs. faulty) and transparency (with vs. without explanation) were manipulated as between-subjects variables, whereas exposure time was used as a repeated-measure variable. A combined qualitative and quantitative methodological approach was used to analyze the data from 171 participants. Our results show that participants perceived the system making a faulty recommendation as a trust violation. Additionally, a trend emerged from both the quantitative and qualitative analyses regarding how the availability of explanations (even when not accessed) increased the perception of a trustworthy system.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Initial Trustmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Trust Development and Explainability: A Longitudinal Study with a Personalized Assistive System

Zafari,

de Pagter,

Papagni

et al. 2024

MTI

View full text Add to dashboard Cite

show abstract

“…Or even when using a trust to seek more control in automated and intelligent system predictions. Or, provide socio-ethical AI as transparency and responsibility solutions through trusted agencies and other audition mechanisms [21,82].…”

Section: The Nature Of Trust Research In Hcimentioning

confidence: 99%

Challenges and Trends in User Trust Discourse in AI Popularity

Sousa

2023

MTI

View full text Add to dashboard Cite

The Internet revolution in 1990, followed by the data-driven and information revolution, has transformed the world as we know it. Nowadays, what seam to be 10 to 20 years ago, a science fiction idea (i.e., machines dominating the world) is seen as possible. This revolution also brought a need for new regulatory practices where user trust and artificial Intelligence (AI) discourse has a central role. This work aims to clarify some misconceptions about user trust in AI discourse and fight the tendency to design vulnerable interactions that lead to further breaches of trust, both real and perceived. Findings illustrate the lack of clarity in understanding user trust and its effects on computer science, especially in measuring user trust characteristics. It argues for clarifying those notions to avoid possible trust gaps and misinterpretations in AI adoption and appropriation.

show abstract

“…Despite the prosperousness of DL at present, how to move forward this technology for the future is still deserved to be dis-cussed. On one hand, continuous efforts on challenging topics, including advanced architecture [5,6], interdisciplinary [7,8], trustable artificial intelligence [9,10], etc., should be made. On the other hand, more works on how to apply DL to real use cases are also necessary, so that this technology will not only be manipulated by scientists or engineers in the field of computer science, but also can be a powerful tool for those who are interested from other fields, and even for all human beings.…”

Section: Introductionmentioning

confidence: 99%

Generalized Operating Procedure for Deep Learning: an Unconstrained Optimal Design Perspective

Chen,

Zhang,

Cui

et al. 2020

Preprint

View full text Add to dashboard Cite

Deep learning (DL) has brought about remarkable breakthrough in processing images, video and speech due to its efficacy in extracting highly abstract representation and learning very complex functions. However, there is seldom operating procedure reported on how to make it for real use cases. In this paper, we intend to address this problem by presenting a generalized operating procedure for DL from the perspective of unconstrained optimal design, which is motivated by a simple intension to remove the barrier of using DL, especially for those scientists or engineers who are new but eager to use it. Our proposed procedure contains seven steps, which are project/problem statement, data collection, architecture design, initialization of parameters, defining loss function, computing optimal parameters, and inference, respectively. Following this procedure, we build a multi-stream end-to-end speaker verification system, in which the input speech utterance is processed by multiple parallel streams within different frequency range, so that the acoustic modeling can be more robust resulting from the diversity of features. Trained with VoxCeleb dataset, our experimental results verify the effectiveness of our proposed operating procedure, and also show that our multi-stream framework outperforms single-stream baseline with 20 % relative reduction in minimum decision cost function (minDCF).

show abstract

Risk and Trust in artificial intelligence technologies: A case study of Autonomous Vehicles

Cited by 16 publications

References 15 publications

Trust Development and Explainability: A Longitudinal Study with a Personalized Assistive System

Trust Development and Explainability: A Longitudinal Study with a Personalized Assistive System

Challenges and Trends in User Trust Discourse in AI Popularity

Generalized Operating Procedure for Deep Learning: an Unconstrained Optimal Design Perspective

Contact Info

Product

Resources

About