Explanations and Expectations

Abstract

Trust is a vital determinant of acceptance of automated vehicles (AVs) and expectations and explanations are often at the heart of any trusting relationship. Once expectations have been violated, explanations are needed to mitigate the damage. This study introduces the importance of timing of explanations in promoting trust in AVs. We present the preliminary results of a within-subjects experimental study involving eight participants exposed to four AV driving conditions (i.e. 32 data points). Preliminary results show a pattern that suggests that explanations provided before the AV takes actions promote more trust than explanations provided afterward.

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

“…For instance, empirical results from a series of studies on explanations by automated vehicles show how the explanations' timing plays a central role in determining users' trust. The researchers found that explanations provided before the vehicle acted had a positive influence on participants' attribution of trust, while explanations that were given after a specific action did not affect trust ratings [41,42]. In line with H1 , these findings suggest that explanations may be more beneficial before, rather than during, an interaction.…”

“…In this regard, several researchers note that, particularly during the first phases of the adoption of and interaction with new AI-based technologies, providing initial explanations may improve trust formation processes. For instance, studies suggest that automated vehicles' explanations provided before the vehicle acted may positively influence the willingness to trust the vehicle [41,42]. Specifically, by compensating for the lack of previous experience (and proven reliability), initial explanations that clarify the functions and purposes of a specific technology may reduce users' perception of risk, increase perceived trustworthiness, enable accountability, and mitigate individual disposition [25,34,43].…”

A Survey of Explainable Artificial Intelligence for Smart Cities

“…In XAI, transportation is an important application area. The explanation of car behavior has even now begun, [137,138], but there is still a long way to go.…”

When to choose ranked area integrals versus integrated gradient for explainable artificial intelligence – a comparison of algorithms

“…According to the literature, there are four reasons for the need for XAI, depending on the motivation (Bojarski et al. , 2016; Haspiel et al. , 2018).…”

“…According to the literature, there are four reasons for the need for XAI, depending on the motivation (Bojarski et al, 2016;Haspiel et al, 2018). The first is to explain or justify, which implies a solution to the problem of unobjective or discriminatory results that presupposes a particular outcome's validity, proving that the algorithms are fair and ethical (Huston, 2021).…”