Abstract: With the advancement of large language models and agent technologies, human-centered computing paradigms are evolving towards deeper collaborative modes. As a frontier research domain, human-AI collaborative computing aims to facilitate the formation of "Human-AI Collaborative Dyad" through tightly coupled cooperation between humans and artificial intelligence (AI), achieving shared cognitive reasoning and collaborative decision-making. Human-AI collaborative decision-making, as a crucial frontier research direction in this field, endeavors to integrate human experiential wisdom with AI's computational capabilities, leveraging their respective strengths to form complementary advantages. This integration significantly enhances efficiency in solving complex problems and promotes decision-making toward higher precision, efficiency, scientific rigor, and flexibility. Such collaborative scenarios exhibit characteristics of interactive learning, dynamic coordination, and continuous evolution between humans and AI, while facing challenges including: difficulties in understanding complex human-AI intentions, challenges in modeling dynamic preferences, interpretability barriers of black-box decision mechanisms, poor controllability of decision processes, and value misalignment in decision outcomes. Focusing on these issues, this paper presents our research team's innovative work in human-AI collaborative decision-making, including intention understanding, personalized modeling, explainability enhancement, and value alignment. Furthermore, we provide profound insights into the future development of human-AI collaborative decision-making systems.