Can AI Plan a Fair Chinese Shared Dinner

Archived study

Builds a Chinese shared-dinner scenario for ten diner profiles and scores menu safety and satisfaction separately. The record includes corrected arithmetic, preference and menu tables, and per-diner validation.

Interpretation and limits

The fairness score depends on a chosen rubric. The archive labels everyone safe while some diners remain unsatisfied; its claim that nobody is disadvantaged should not be read as a universal fairness or allergy-safety guarantee.

Package and provenance

The original .science export and matching poster are preserved byte-for-byte from the upstream case. The package contains the conversation, session metadata, and the resources inventoried in its internal manifest. This introduction summarizes the archived record; the research, external sources, and model evaluations were not rerun for this import.