Can AI truly understand the world around us? One of the biggest hurdles for artificial intelligence is grasping common sense—knowing that "birds fly" is plausible while "airplanes swim" is not. Researchers are tackling this challenge by giving AI a knowledge boost. In a new study presented at the MSP2023 conference, scientists explored how to enhance an AI's understanding of semantic plausibility, which essentially means figuring out if an event makes sense in the real world. They focused on injecting external knowledge about entities (like people, places, and things) and events (like actions or occurrences) into a large language model (LLM). Think of it like teaching the AI the definitions of words and how they relate to each other. The researchers used a clever method: they combined information from a knowledge base with carefully designed templates to feed the LLM natural language prompts. For example, they might tell the AI that a 'bird' is a 'feathered animal that can fly,' and then ask it to judge the plausibility of 'a bird soaring through the sky.' They also used a technique called data augmentation to ensure the AI saw a balanced mix of plausible and implausible events, which is crucial for effective learning. The results? The AI’s performance in judging the plausibility of simple events improved significantly, demonstrating the power of external knowledge. However, challenges remain, especially when dealing with more abstract or complex events. For instance, the AI might struggle to differentiate the plausibility of “a company launching a product” vs. “a cat launching a product.” This research takes a promising step towards imbuing AI systems with a much-needed dose of common sense. By understanding entities, events, and their relationships, AI can move closer to genuinely comprehending our world and interacting with us more naturally.