Evaluating the quality of machine-generated text has always been a challenge. How do you teach a computer to understand nuance, creativity, and the subtle art of human language? Researchers are tackling this problem with innovative new approaches, and CHECK-EVAL is one of the most promising. Imagine having a meticulous editor reviewing every piece of written content, carefully checking for key elements and ensuring it meets specific quality standards. That's the essence of CHECK-EVAL, a framework that uses AI to evaluate AI-generated text. Instead of relying on traditional metrics that often miss the mark, CHECK-EVAL utilizes checklists tailored to different evaluation criteria. These checklists guide the evaluation process, focusing on crucial aspects like factual accuracy, relevance, coherence, and even fluency. This method employs Large Language Models (LLMs) in two stages: checklist generation and evaluation. First, the LLM creates a checklist of essential points based on the source material. Then, it meticulously compares the generated text to this checklist, providing a detailed and interpretable assessment. What sets CHECK-EVAL apart is its structured and interpretable nature. Unlike traditional metrics that offer a single, often opaque score, CHECK-EVAL pinpoints specific strengths and weaknesses. This targeted feedback is invaluable for developers, helping them refine their models and generate higher-quality text. Early testing on datasets like the Portuguese Legal Semantic Textual Similarity dataset and the SUMMEVAL dataset is encouraging. CHECK-EVAL consistently outperforms existing metrics, showing a stronger correlation with human judgment. While CHECK-EVAL relies on the capabilities of existing LLMs, which have their own limitations, it represents a significant step forward in automated text evaluation. As AI-generated text becomes more prevalent, tools like CHECK-EVAL are crucial for ensuring quality and building trust in machine-written content. Future research will focus on refining the checklist generation process, extending CHECK-EVAL to different writing tasks, and optimizing its efficiency. The potential for this framework to revolutionize how we evaluate text quality is immense, paving the way for more reliable and impactful AI-driven communication.