Transparency in AI is crucial for fostering user trust and acceptance, yet achieving it through explanations presents significant design challenges, particularly regarding how much detail to provide. For example, in-depth explanations can convey accurate and comprehensive information, but they also risk overwhelming users. This paper considers this important design tradeoff in the context of training dataset explanations, which describe the data used to train AI systems and differ from most model-centric explanations in terms of what and how much information they communicate. Specifically, we investigate how information depth in training dataset explanations and the use of Progressive Disclosure impact users’ understanding of an AI system (assessed via their critiques of the system), their system assessments, and their cognitive load. Findings from a study with 32 participants show advantages to providing users with comprehensive information on training datasets. Detailed explanations not only enhanced perceived trust, fairness, and understanding, but were also preferred by participants despite the increased cognitive load. While Progressive Disclosure did not effectively mitigate cognitive load, it improved users’ perception of learning. These findings suggest that effective transparency does not come from minimizing detail, but from embracing it, as participants consistently valued clarity and completeness over brevity, even at the cost of higher cognitive load.

The widespread use of digital devices among children and teenagers has raised concerns about overuse, particularly for early adolescents, who have unique developmental needs and engage with technology more frequently than other age groups. A challenge for designers and researchers interested in contributing solutions is a lack of synthesized design guidelines and characterization of the current state-of-the-art. In this paper, we present a systematic scoping review of academic literature and an analysis of 47 apps, providing a comprehensive characterization of existing tech-mediated solutions for early adolescents. Our review covers literature from two major databases (ACM DL and IEEE Xplore) spanning the past 10 years (2014-May 2024), following the scope of prior similar reviews. The app analysis includes Google Play and Apple App Store apps with features targeting tech overuse, excluding general-purpose apps (e.g., social media, games) and apps without a free trial version. Our findings highlight researchers' design recommendations for promoting tech disengagement in this demographic (e.g., supporting collaborative rule-setting and self-monitoring, maintaining privacy, addressing diverse user needs), while revealing that existing apps tend to prioritize restrictive measures, overlooking self-regulation and active parental engagement. Our findings also identify areas of agreement and potential misalignments between current digital interventions and prior research on target users' preferences.

Training data has a profound impact on the performance of Machine Learning (ML) systems. To help interested parties develop informed trust and appropriate reliance, recent work has proposed providing users with training dataset explanations. While this prior work showed promising subjective impacts of increasing system transparency in this manner, little is known as to how users might use this information to critique a system. In this work, we investigate how two presentation styles for such explanations (a narrative-driven Data Story and a Q&A format) support users’ critique of an automated hiring system. Findings from a between-subjects study with 39 participants provide insights into the aspects of training dataset explanations that participants leveraged in their critiques. Our findings also show that presentation style can impact critique emphasis, critique accuracy, and subjective impressions of explanation utility.

Recruiting participants is a pivotal yet challenging aspect of Human-Computer Interaction (HCI) research. In this position paper, we reflect on our experiences recruiting participants for multiple user studies within a human-centered Explainable AI project. Emphasizing the importance of participant diversity in terms of their AI knowledge and experiences, we discuss challenges, trade-offs, and strategies associated with recruiting for both in-person and remote participation.

Training datasets fundamentally impact the performance of machine learning (ML) systems. Any biases introduced during training (implicit or explicit) are often reflected in the system's behaviors leading to questions about fairness and loss of trust in the system. Yet, information on training data is rarely communicated to stakeholders. In this work, we explore the concept of data-centric explanations for ML systems that describe the training data to end-users. Through a formative study, we investigate the potential utility of such an approach, including the information about training data that participants find most compelling. In a second study, we investigate reactions to our explanations across four different system scenarios. Our results suggest that data-centric explanations have the potential to impact how users judge the trustworthiness of a system and to assist users in assessing fairness. We discuss the implications of our findings for designing explanations to support users’ perceptions of ML systems.