A Model for Artificial Conscience to Control Artificial Intelligence

Abstract

We propose “Artificial Conscience - Control Module” framework to control the AI systems and to make them adaptable based on the user requirements. AI users can have different needs from the same AI system, and these systems must adjust their output based on these requirements. The proposed framework enables users to provide context to the AI system by assigning weights to different evaluation metrics. Based on these weights, AI metrics-agents negotiate with each other using our trust engine to output a solution with maximum “Artificial Feeling.” This framework can be easily implemented in any AI system where multiple metrics are involved. We have illustrated the proposed framework using an AI system for classifying people based on income.