Teachable Agent

Teachable agents are AI systems designed to learn from human instruction, aiming to improve both the AI's capabilities and the human instructor's understanding. Current research focuses on enhancing these agents' ability to handle diverse tasks and complex interactions across various modalities, including visual and language-based instruction, often leveraging large language models and memory-augmented architectures. This research is significant for advancing AI's ability to learn and interact naturally with humans, with potential applications in education, robotics, and personalized learning environments. The impact of factors like misrecognition and conversational rapport on learning outcomes is also a key area of investigation.