AI Hallucination

The phenomenon where an AI model generates confident, plausible-sounding responses that are factually incorrect, fabricated, or not grounded in the actual input data. In accessibility contexts, AI hallucinations pose a serious safety concern — for example, a multimodal AI confidently stating that an elevator is located behind trash bins when it is not, or providing incorrect bus route numbers. For blind users who rely on AI for navigation and environmental understanding, hallucinations can lead to disorientation, wasted time, and potentially dangerous situations.