Ethical AI

Ethical AI research aims to develop and deploy artificial intelligence systems that align with human values and avoid causing harm. Current efforts focus on mitigating biases in datasets and models (like diffusion models and large language models), enhancing explainability and interpretability through techniques such as Chain-of-Thought reasoning and mathematical certification of properties like partial monotonicity, and addressing uncertainty in AI decision-making. This field is crucial for ensuring responsible AI development and deployment across various sectors, impacting not only the trustworthiness of AI systems but also their societal impact and legal frameworks governing their use.