Imagine grading essays not one by one, but by comparing them in pairs. This intriguing concept is at the heart of a new research paper that explores how Large Language Models (LLMs) can revolutionize automated assessment. Traditionally, LLMs have been used to score assignments individually, but this approach has limitations. This new research delves into 'comparative assessment,' where LLMs judge which of two texts is superior based on specific criteria like coherence or complexity. The fascinating twist? This method taps into the subtle art of comparison, mirroring how humans often evaluate quality. By fine-tuning LLMs on comparisons rather than absolute scores, researchers found a clever way to boost efficiency. This method streamlines the evaluation process, significantly reducing the computational workload without sacrificing accuracy. The potential impact is huge, especially for tasks like grading large-scale exams or evaluating responses in educational settings. While the research uses smaller models like GPT4o-mini and Llama-3.1-8B due to computational constraints, the findings pave the way for more robust, efficient assessment methods in the future. This innovative approach promises to not only expedite automated evaluation but also to enhance its nuance and effectiveness, bringing us closer to truly intelligent assessment systems.