Artificial intelligence (AI) systems have revolutionized many fields, but they are often criticized for lacking ethical reasoning. We see this reflected in news stories about biased algorithms or AI-generated content that violates social norms. Addressing this issue is crucial for responsible AI development, but how can we teach machines about ethics? One intriguing approach focuses on aligning language models with human-like reasoning, ensuring they not only perform tasks accurately but also make morally sound decisions. Researchers are exploring how to inject ethical considerations into AI models, and a recent paper introduces a novel method centered around the 'Dataset for Aligning Reasons' (DFAR). This dataset contains statements labeled as ethical or unethical, along with human-provided explanations for these judgments. The idea is to fine-tune large language models (LLMs) on this data, so they learn to generate reasons for ethical or unethical classifications similar to how humans do. Researchers experimented with two popular LLMs, Llama-2 and Mistral, fine-tuning them on DFAR. The results were impressive. The fine-tuned models not only excelled in classifying ethical and unethical statements but also generated explanations that closely aligned with human reasoning. This suggests that feeding AI with structured, human-like rationales can enhance their ethical decision-making. The study's key innovation lies in this combination of labels (ethical/unethical) with corresponding reasons, a technique not typically used in previous alignment studies. This approach tackles two issues simultaneously: correctly classifying scenarios and providing strong human-like reasons behind the classifications. The research also acknowledges that the journey towards perfectly ethical AI is ongoing. Even with improved performance, some misalignments persist, particularly in statements with ambiguous contexts. Further data refinement and exploration of more advanced NLP techniques might be necessary to fine-tune ethical AI even more. Nevertheless, the research offers a promising step towards aligning AI with human values and building a more trustworthy and morally responsible future for AI.