BaZi-Based Character Simulation Benchmark: Evaluating AI on Temporal and Persona Reasoning
Siyuan Zheng ∗♣△ , Pai Liu ∗†♣♠ , Xi Chen ∗♣♠ , Jizheng Dong ♡ , Sihan Jia ♢ · 2025
Read BaZi-Based Character Simulation Benchmark: Evaluating AI on Temporal and Persona Reasoning by Siyuan Zheng ∗♣△ , Pai Liu ∗†♣♠ , Xi Chen ∗♣♠ , Jizheng Dong ♡ , Sihan Jia ♢ (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-23337.txt.
Description
Siyuan Zheng and four co-authors build a benchmark for AI character simulation based on BaZi, a traditional Chinese system that reads life patterns from birth time. Virtual characters for games and virtual reality need coherent personas, yet current methods lean on costly annotation or handcrafted prompts. Their data set turns real experiences into life-event questions across wealth, health, kinship, career, and relationships.
They also propose a BaZi-LLM system that joins symbolic reasoning with a language model to generate personas over time. The benchmark tests temporal and persona reasoning together, a fresh angle on consistent character AI.
Catalog record
| CREATOR | Siyuan Zheng ∗♣△ , Pai Liu ∗†♣♠ , Xi Chen ∗♣♠ , Jizheng Dong ♡ , Sihan Jia ♢ |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-23337