{
  "id": 11771,
  "url": "https://arxiv.org/abs/2607.15740v1",
  "title": "Debiasing Text-to-Image Evaluation via Implicit Cultural Alignment Reward Modeling",
  "summary": "As Text-to-Image (T2I) systems rapidly advance, evaluating the cultural authenticity of synthesized content has become increasingly important for fair and trustworthy generative AI. Existing T2I evaluation metrics and multimodal judges often rely on visual-semantic representations that underrepresent implicit cultural norms, leading to biased preference judgments and the omission of fine-grained cultural cues. In addition, visual question answering (VQA)-based evaluators typically depend on auto",
  "authors": "Bo-An Chang, Yu-Chih Chen",
  "category": "research",
  "topics": "bias-fairness,safety-alignment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-07-17T08:23:03.000Z",
  "fetched_at": "2026-07-20T05:10:09.534Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/11771",
  "original_url": "https://arxiv.org/abs/2607.15740v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}