Trust in the Machine: The Impact of Virtual Agent Anthropomorphism on User Trust and Compliance in Circular Ecosystems.

Abstract

The growing global collection of hazardous electronic waste, especially disposed consumer lithium-ion batteries, has developed a serious environmental crisis. A major shift toward a closed-loop circular economy from a traditional linear "take-make-dispose" economy is necessary in this situation. The usage of digital systems is rapidly growing to coordinate waste collection and to manage the safety rules required for harmful materials under the new concept of Circular Economy 4.0. This study investigates the influence of digital interface design in motivating consumer participation in recycling programs for hazardous materials. Especially, the study examines whether the anthropomorphic (human-like) virtual avatar design assists users to follow complex, safety-critical guidelines in absence of human experts to help them.

The research builds on a theoretical model that incorporates the Elaboration Likelihood Model (ELM), Perceived Risk Theory, and the “Computers are Social Actors” (CASA) model. This model tests the interaction of different types of virtual agents (human-like vs machine-like) with a user's perceived efficiency and perceived greenwashing. The ultimate objective of the structural model is to predict a user's actual behavioural intention to comply with safety rules. A quantitative experiment was performed using Partial Least Squares Structural Equation Modelling (PLS-SEM) to analyse collected data to evaluate the structural model. To ensure the accuracy and robustness of the final results across the population, demographic factors were also added as control variable.

The analysis of the experimental data showed that the avatar's visual design had no direct influence on a user's behavioural intention to follow guidelines for recycling hazardous materials. The change of visual also failed to activate any of the hypothesized cognitive mediators. This lack of influence is explained as an "Avatar-Blindness" paradox. It happens when an anthropomorphic (human-like) digital avatar fails to produce expected social cues in safety-critical environment. Instead, the findings of the study show that users compliance behaviour is particularly guided by central route of cognitive processing. This route is defined by an in-depth evaluation of the system's functional usefulness and a critical check of the corporate integrity of green claims. The study concludes that users ignore aesthetic cues and focus on functional usefulness and transparent information when they encounter significant material risks. The success of digital systems in circular economy does not depend on visual aesthetic. Instead, the success depends on developing tools that actually fit the task, work effectively, and maintain ethical transparency.