Abstract

Artificial intelligence (AI) is increasingly used in organizational and everyday decision-making, but its effectiveness depends on appropriate user trust. Both under-trust and over-trust are problematic because users may reject useful AI systems when trust is too low or over-rely on imperfect systems when trust is too high (Lee & See, 2004). This TREO paper develops a conceptual model based on the theory of reasoned action (TRA) to explain how anthropomorphism and explainability influence trust through perceived competence, benevolence, and integrity (Ajzen & Fishbein, 1980). Anthropomorphic cues may increase perceived competence and benevolence by making AI appear more relatable and capable (Epley et al., 2007). However, excessive or inauthentic human likeness may undermine perceived integrity by creating unrealistic expectations or suspicion (Glikson & Woolley, 2020). Explainability strengthens trust by reducing uncertainty and helping users understand, evaluate, and contest AI decisions (Miller, 2019). This TREO paper proposes a conceptual model examining how anthropomorphism and explainability may influence user trust in AI through perceptions of competence, benevolence, and integrity. Anthropomorphism is expected to increase perceived competence and benevolence, while excessive human-likeness may reduce perceived integrity. Explainability is expected to strengthen all three trustworthiness perceptions, which in turn are expected to shape trust in AI. This study contributes to the AI trust research by integrating anthropomorphism, explainability, and trustworthiness beliefs into a unified TRA-based framework. In practice, it suggests that designers should balance relatable AI interactions with transparency, ethical accountability, and calibrated reliance. Future researchers can test the model using survey data and structural equation modeling across AI application domains such as healthcare, education, finance, and workplace decision support.

Download

Share

COinS