Self Awareness

Self-awareness in artificial intelligence focuses on developing systems that can understand and report their own internal states, knowledge limitations, and decision-making processes, ultimately aiming to improve reliability and trustworthiness. Current research emphasizes enhancing self-awareness in large language models (LLMs) and multimodal models through techniques like multi-round reasoning, contextualized attention mechanisms, and reinforcement learning from knowledge feedback, often incorporating hierarchical scene graphs or heterogeneous graph neural networks for improved representation and reasoning. This research is crucial for mitigating issues like hallucinations and biases in AI systems, leading to more robust and reliable applications across diverse fields, including robotics, medical image analysis, and human-computer interaction.