A conceptual architecture of sincerity in software agents for task performance

The advent of software agent technology manifests many benefits to humans in many aspects; especially in alleviating the burden of humans’ tasks. However, the technology entails the emergence of self-interested behaviour of agents that can create negative impacts if ethical elements are not designed in the architecture of agent-based applications. Consequently, researchers in autonomous systems technology are studying human’s strong moral values to be implemented in the technology as a means to avoid unnecessary problems in task performance as a consequence of the self-interested behaviour. Sincerity is one human’s strong moral value that is suitable to be implemented, especially for conflict-prone tasks. A software agent architecture for instilling sincerity behaviour should be designed to implement such behaviour in multi-agent environments. To do this, a survey and observation of related humans’ activities in completing tasks have been conducted to understand the factors of human sincerity while performing the tasks. Correspondingly, software agents and their characteristics are investigated to discover the clues for instilling sincerity behaviour in software agents. In this paper, we reconcile these two aspects to propose a conceptual software agent architecture for implementing sincerity behaviour while agents are performing their tasks.