Influence of self‐concept clarity on advice seeking and utilisation

Co-design of human-centered, explainable AI for clinical decision support

eXplainable AI (XAI) involves two intertwined but separate challenges: the development of techniques to extract explanations from black-box AI models, and the way such explanations are presented to users, i.e., the explanation user interface. Despite its importance, the second aspect has received limited attention so far in the literature. Effective AI explanation interfaces are fundamental for allowing human decision-makers to take advantage and oversee high-risk AI systems effectively. Following an iterative design approach, we present the first cycle of prototyping-testing-redesigning of an explainable AI technique, and its explanation user interface for clinical Decision Support Systems (DSS). We first present an XAI technique that meets the technical requirements of the healthcare domain: sequential, ontology-linked patient data, and multi-label classification tasks. We demonstrate its applicability to explain a clinical DSS, and we design a first prototype of an explanation user interface. Next, we test such a prototype with healthcare providers and collect their feedback, with a two-fold outcome: first, we obtain evidence that explanations increase users’ trust in the XAI system, and second, we obtain useful insights on the perceived deficiencies of their interaction with the system, so that we can re-design a better, more human-centered explanation interface.

A meta-analysis of the weight of advice in decision-making

The degree to which people take advice, and the factors that influence advice-taking, are of broad interest to laypersons, professionals, and policy-makers. This meta-analysis on 346 effect sizes from 129 independent datasets (N = 17, 296) assessed the weight of advice in the judge-advisor system paradigm, as well as the influence of sample and task characteristics. Information about the advisor(s) that is suggestive of advice quality was the only unique predictor of the overall pooled weight of advice. Individuals adjusted estimates by 32%, 37%, and 48% in response to advisors described in ways that suggest low, neutral, or high quality advice, respectively. This indicates that the benefits of compromise and averaging may be lost if accurate advice is perceived to be low quality, or too much weight is given to inaccurate advice that is perceived to be high quality. When examining the three levels of perceived quality separately, advice-taking was greater for subjective and uncertain estimates, relative to objective estimates, when information about the advisor was neutral in terms of advice quality. Sample characteristics had no effect on advice-taking, thus providing no evidence that age, gender, or individualism influence the weight of advice. The findings contribute to current theoretical debates and provide direction for future research.