AI hallucination happens when an AI model makes up information. Instead of saying “I don’t know,” it delivers a confident, yet incorrect, response. This is a major issue in grading since fabricated details can mislead students and erode trust. Ednius tackles this by using multiple specialized AI agents that check each other’s work, ensuring every grade is accurate and based on real data.