Can Artificial Intelligence Be the Key to Universal Ethics?

Today, advances in artificial intelligence (AI) technologies hold the potential to provide solutions to many challenges facing humanity. However, whether this technology can serve as a universal moral guide—particularly in ethical and social matters—is a subject of debate. Human morality has historically been shaped by religions, cultures, emotions, and social norms, exhibiting a relative and variable structure.

Can an AI system, freed from emotional biases and historical burdens, enable humanity to find common moral ground?

Moral decisions are not solely the product of rational processes; they are deeply intertwined with emotions, social conditions, and individual interests. For instance, psychological studies have shown that judges’ decisions can be influenced by physiological factors such as fatigue or hunger (Danziger et al., 2011). This reveals the subjectivity inherent in human decision-making processes.

AI, by contrast, can make consistent and data-driven decisions independent of such biases. For example, autonomous vehicles may respond to accident scenarios by calculating the option that minimizes harm, thereby avoiding emotional reactions. While this suggests AI could reduce human errors, it is essential not to overlook the fact that moral decisions cannot be reduced solely to mathematical optimization.

Establishing a universal moral framework is extremely difficult due to conflicts between the values of different cultures. For instance, the balance between freedom of expression and hate speech is interpreted differently in Western societies compared to Eastern ones. When AI is programmed according to a specific set of values, the critical question arises: who determines these values and by what criteria?

Bostrom and Yudkowsky (2014) argue that AI’s moral decision-making mechanisms must be aligned with human values. Yet the question “whose human values?” remains unanswered. For example, an AI system adopting a utilitarian approach might maximize “the happiness of the majority” while disregarding the rights of minorities.

AI systems are limited by the datasets and algorithms upon which they are trained. As a result, human biases can be reflected in these datasets (Bolukbasi et al., 2016). For instance, an AI model used in hiring processes may learn and perpetuate gender discrimination present in historical data.

To address this issue, transparency and oversight mechanisms must be integrated into AI decision-making processes. Furthermore, fundamental values such as universal human rights principles can serve as a framework for AI training. However, since even the universal acceptance of these values is contested, AI currently cannot be regarded as a fully autonomous moral agent.

While AI cannot fully resolve the complexity of human morality, it can offer more consistent and transparent decision-making processes. Yet, creating a universal moral system is not merely a technological challenge—it requires philosophical, cultural, and political consensus. AI may serve as a guide in this process, but final decisions must remain within the realm of humanity’s shared moral dialogue.