Decoding Ethics: Proficiency of LLMs in Addressing Moral Dilemmas

In the dynamic global digital landscape, the mainstream use of Artificial Intelligence led us into a transformative era of technical ubiquity. Pioneering this movement, are advanced Artificial Intelligence models like LLMs. Large Language Models (LLMs) are Machine Learning/Deep Learning models that show remarkable abilities to generate human-like text. They are trained on vast amounts of data as a result of which they imbibe the nuances of human conversation up to great accuracies. They are mainly used for content generation, along with other natural language processing tasks. From medicine to education, LLMs find applications in nearly all fields of society. However, several ethical and moral implications need to be taken into account to ensure the responsible use of LLMs. Our work explores the performance of ChatGPT-3.5 Turbo, a text-generative Large Language Model, in the understanding and comprehension of questions entailing moral and ethical dilemmas using accuracy as the key evaluation metric. Ethical considerations in terms of biases, transparency, reliability, and accuracy need to be analyzed to ensure the responsible use of LLM technology. There is currently a limited understanding of LLM reasoning processes and ability of handle real-world, ethical dilemmas. Through this work, we aim to assess the decision-making capabilities of LLMs in terms of moral scenarios and contribute to the establishment of ethical norms for the standardized use of Large Language Models, making them more secure and accessible.