Optical Illusion

Optical illusions are perceptual discrepancies between physical reality and subjective experience, serving as a crucial tool for understanding the human visual system and limitations of artificial intelligence. Current research focuses on leveraging deep learning models, particularly diffusion models, to generate and analyze illusions, investigating their robustness and susceptibility to adversarial attacks across various modalities (image, text, audio). This work has implications for improving AI systems' perceptual accuracy, enhancing the design of robust autonomous systems (e.g., self-driving cars), and furthering our understanding of human visual perception and cognitive biases.