{
  "id": 953,
  "url": "https://arxiv.org/abs/2606.16799v1",
  "title": "Decoupling Semantics from Distortions: Multi-Scale Two-Stream Vision-Language Alignment for AI-Generated Image Quality Assessment",
  "summary": "Existing vision-language model (VLM)-based AI-generated image quality assessment (AIGIQA) methods suffer from a fundamental semantic-distortion dimensional conflict: monolithic representations optimized for semantic discrimination inherently entangle compositional understanding with low-level perceptual sensitivity, rendering them blind to fine-grained quality degradations. We introduce MST-CLIPIQA, a multi-scale two-stream framework that achieves hierarchical vision-language alignment through e",
  "authors": "Zijie Meng",
  "category": "research",
  "topics": "bias-fairness,safety-alignment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-06-15T14:40:30.000Z",
  "fetched_at": "2026-07-14T14:14:54.533Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/953",
  "original_url": "https://arxiv.org/abs/2606.16799v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}