Black Box Large Language Model

Black-box large language models (LLMs) are powerful AI systems whose internal workings are opaque, posing challenges for understanding their behavior and improving their performance. Current research focuses on methods to enhance their capabilities without direct access to their internal parameters, including techniques like prompt engineering, adapter models, and uncertainty quantification methods to assess reliability. These efforts aim to improve LLM accuracy, trustworthiness, and efficiency across diverse applications such as question answering, recommendation systems, and text generation, while mitigating risks associated with bias and adversarial attacks.