Xiong, Guojun, and Milind Tambe. “VORTEX: Aligning Task Utility and Human Preferences Through LLM-Guided Reward Shaping”. Proceedings of the AAAI Conference on Artificial Intelligence, vol. 40, no. 32, Mar. 2026, pp. 27162-70, doi:10.1609/aaai.v40i32.39931.