Can AI truly understand cause and effect? Researchers explored this question by investigating whether Large Language Models (LLMs) can automatically verify causal relationships described in text. Imagine a causal graph, a visual representation of how different variables influence each other (like smoking causing lung cancer). Traditionally, verifying these graphs requires expert evaluation, a time-consuming and expensive process. This research aimed to automate this verification using LLMs like BERT and ChatGPT. Two main approaches were tested: fine-tuning pre-trained LLMs on labeled data specific to causal relations and prompting LLMs with carefully crafted instructions and examples. Surprisingly, the results showed that fine-tuned models significantly outperformed prompt-based methods, achieving up to a 20.5% higher F1 score in correctly classifying causal relationships. This contrasts with recent trends where prompt-based LLMs have shown impressive performance in various tasks. One reason for this discrepancy might be the implicit way causality is often expressed in language. While humans can infer causal links from phrases like "contributes to" or "plays a role in," LLMs struggle without explicit causal cues like "causes" or "caused by." Fine-tuned models, having learned from labeled examples, are better equipped to recognize these implicit patterns. Another interesting finding was the importance of context. Providing the LLM with the surrounding text of the target entities improved performance, suggesting that LLMs benefit from a broader understanding of the situation. While fine-tuning proved more effective in this study, it requires extensive labeled data, which can be a bottleneck. Prompt-based methods, though currently less accurate, offer a more flexible and potentially scalable approach. Future research will focus on improving prompt engineering techniques, perhaps by incorporating "chain-of-thought" prompting or providing LLMs with additional background knowledge. This research highlights the challenges of teaching AI to understand causality, a fundamental aspect of human reasoning. As LLMs evolve, the ability to automatically verify causal claims could revolutionize fields like medicine, economics, and social sciences, enabling faster and more reliable insights from complex data.