Imagine an AI that can understand medical jargon in multiple languages, helping doctors stage lung cancer more efficiently. That's the tantalizing possibility explored by researchers using large language models (LLMs). A recent study investigated how these powerful AIs perform when tasked with classifying lung cancer stages from radiology reports in both English and Japanese. The task is tricky: radiologists write reports in a narrative style, and important staging information (TNM classification – Tumor, Node, Metastasis) is embedded within the text. This research used GPT-3.5-turbo, a multilingual LLM, to automatically extract this crucial data. Surprisingly, the study found that these models can do a decent job of staging cancer even without special training, especially when given clear definitions of the TNM stages. Accuracy was highest when the reports and definitions were in English, correctly identifying the metastasis stage (M) in a whopping 94% of cases. Accuracy dipped slightly for Japanese reports, highlighting the challenges LLMs still face with languages other than English. The study also revealed that giving the LLM the full definition of each stage (T, N, and M) boosted accuracy considerably. This suggests that while LLMs possess some base medical knowledge, carefully crafted prompts can unlock their full potential. While this research is a significant first step, challenges remain. The dataset was relatively small, and the use of translated texts might have skewed the results. However, these initial findings are incredibly exciting. Imagine a future where multilingual AI assistants rapidly analyze medical images, extract key findings in any language, and offer insights to oncologists worldwide. This technology could transform cancer care and ensure that every patient, regardless of where they live, has access to the best possible diagnosis and treatment.