Reproduce “Auditing World-Model Explanations in Retail Forecasting: Faithfulness, Accuracy and Price Response” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Karthik Iyer

The authors audit whether retail models’ promotion-lift explanations reflect what promotions cause. They compare stated lift with known effects in generated retail data, changing whether plans favor high-demand weeks. In that setting, world-model lift totals 1.33–1.40 times truth, despite passing several checks; this is a simulation result, not evidence of real-world loss or prevalence.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.