Ethics in the Absence of Oversight: Evaluating AI's Moral Reasoning

Artificial Intelligence (AI) systems are becoming more prominent in contemporary times and continue to grow rapidly. With the rise of autonomous AI, ethical and philosophical concerns have also increased. When the boundaries of human control begin to blur, it may create an “ethical vacuum.” Current research has discussed this issue theoretically, but very few studies have tested it practically. Decision-making in the absence of human guidance does not ensure fairness, empathy, or accountability, which shows that AI’s ethical reasoning shifts when oversight is removed. Although AI models can solve complex problems and reduce human errors, they often lack moral reasoning. This study aims to analyze how moral reasoning differs when human control is explicitly given or completely removed. The paper also discusses different forms of human control—direct, indirect, and proportional—to build the theoretical foundation of the study. The study consists of two main components. First, a chatbot experiment will be conducted using moral dilemmas related to fairness, responsibility, and empathy. Each scenario will be tested under two conditions:1. without any moral guidance, allowing AI to respond autonomously and 2. with a human-guided prompt containing ethical instructions focusing on fairness and accountability. The collected responses will be evaluated using a rubric. Second, a short survey will be conducted to understand which responses appear more ethical, trustworthy, and aligned with moral values. Expected findings suggest that responses generated without human instructions will be less empathetic and may lack moral nuance. In contrast, responses created with ethical guidance are expected to show higher ethical alignment. Human participants are likely to prefer guided responses, reinforcing the need for human oversight. The final section will present the overall findings and outcomes of the study.