Imagine a future where your car could "see" around corners by having a conversation with other vehicles. This isn't science fiction, but a real possibility being explored by researchers using large language models (LLMs). Traditionally, extending a car's field of view relied on vehicle-to-infrastructure communication or sharing sensor data between vehicles, both with their drawbacks. Researchers are now experimenting with LLMs, the technology behind ChatGPT, as a communication interface between vehicles. The idea? Cars equipped with LLMs could exchange textual descriptions of what they "see", allowing an ego vehicle to tap into the perspective of other vehicles, effectively peering around corners. Initial tests with GPT-4V and GPT-4o show promising results. These LLMs can accurately describe traffic scenes and identify pedestrians, even providing details about their position and direction. This granular understanding of the surroundings could significantly improve safety, especially in occluded environments. However, while these LLMs excel at describing what they see in natural language, accurately pinpointing the location of objects using coordinates remains a challenge. There's inconsistency across different attempts and a wide variability in performance depending on the image. Although current LLMs aren't ready to replace dedicated perception systems, they demonstrate the potential for a future where intelligent vehicles communicate with each other, creating a shared understanding of the road and improving safety for everyone.