International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 13 Issue: 02 | Feb 2026
p-ISSN: 2395-0072
www.irjet.net
A REVIEW OF DYNAMIC TRUST-AWARE EXPLAINABILITY MODEL FOR ARTIFICIAL INTELLIGENCE SYSTEMS OPERATING IN HIGHCONSEQUENCE DECISION DOMAINS Sapna Singh1, Mrs. Arifa Khan2 1Master of Technology, Computer Science and Engineering, Lucknow Institute of Technology, Lucknow, India 2Assistant Professor, Department of Computer Science and Engineering, Lucknow Institute of Technology,
Lucknow, India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Artificial Intelligence (AI) systems are
responsible AI deployment (Gunning, 2017; Doshi-Velez and Kim, 2017). This section contextualizes the review by examining the operational landscape of AI in critical domains, the theoretical need for dynamic trust-aware explainability, and the methodological approach adopted for literature synthesis.
increasingly deployed in high-consequence decision domains such as healthcare, autonomous transportation, finance, and defense, where erroneous or opaque decisions can result in severe societal, ethical, and economic impacts. In such environments, establishing calibrated human trust and ensuring model explainability are critical requirements. While substantial progress has been made in Explainable AI (XAI) and computational trust modeling independently, the integration of dynamic trust mechanisms with adaptive explainability remains fragmented across the literature. This review systematically analyzes existing approaches to trustaware explainability, focusing on models that dynamically adjust explanations based on contextual risk, user expertise, and system performance. We categorize current research into intrinsic and post-hoc explainability methods, static and dynamic trust estimation frameworks, and integrated trust– explainability architectures. Furthermore, we evaluate domain-specific implementations and comparative evaluation metrics used to assess explanation quality and trust calibration. The review identifies key research gaps, including the absence of standardized benchmarks, limited real-time adaptability, and insufficient human-centered validation. Finally, we outline future research directions aimed at developing unified, context-sensitive, and regulatorycompliant trust-aware explainable AI systems for safetycritical applications.
1.1 Background 1.1.1 AI in Critical Sectors AI technologies are now widely deployed in healthcare for diagnostic support and treatment planning, in autonomous vehicles for perception and navigation, in defense for surveillance and threat assessment, and in finance for credit scoring and fraud detection. In healthcare, deep learning models have achieved expert-level performance in medical imaging tasks, yet their opacity raises concerns about clinical accountability (Esteva et al., 2017). Similarly, autonomous driving systems rely on complex perception pipelines where failure can result in catastrophic consequences, highlighting the necessity for interpretable decision pathways (Bonnefon, Shariff and Rahwan, 2016). In finance, algorithmic decisionmaking affects credit eligibility and risk profiling, often raising fairness and transparency issues (Barocas and Selbst, 2016). Across these sectors, AI systems operate under regulatory scrutiny and ethical constraints, reinforcing the demand for explainable and trustworthy models.
Key Words: Explainable Artificial Intelligence (XAI), Dynamic Trust Modelin, Trust-Aware Systems, HighConsequence Decision Domains, Human–AI Interaction, AI Accountability
1.1.2 Importance of Trust and Explainability Trust in AI systems is a multidimensional construct involving reliability, predictability, and perceived competence. Without appropriate explanations, users may either over-trust automated systems (automation bias) or under-trust them (algorithm aversion), both of which degrade decision quality (Lee and See, 2004; Dietvorst, Simmons and Massey, 2015). Explainability mechanisms aim to render model reasoning comprehensible, thereby supporting trust calibration rather than blind reliance. The DARPA XAI initiative formalized this need by emphasizing that AI systems must provide human-understandable justifications to support operational decision-making (Gunning, 2017). Consequently, trust and explainability are interdependent constructs in high-consequence domains.
1. INTRODUCTION Artificial Intelligence (AI) systems are increasingly embedded in socio-technical infrastructures where their decisions directly influence human safety, legal standing, and economic stability. In high-impact environments, performance accuracy alone is insufficient; systems must also provide intelligible reasoning and maintain calibrated trust relationships with users. The emergence of Explainable Artificial Intelligence (XAI) and computational trust modeling reflects the recognition that transparency, accountability, and reliability are foundational for
© 2026, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 592