{
  "id": 3855,
  "url": "https://arxiv.org/abs/2605.23819v1",
  "title": "Not Too Generative, Not Too Discriminative: The Human Alignment Sweet Spot",
  "summary": "A central question in computational vision is whether human-like visual representations are better explained by discriminative or generative learning. Existing comparisons, however, often confound the learning objective with architecture, scale, and training data, leaving open whether the objective itself drives alignment. We address this confound using Joint Energy-Based Models (JEMs), which interpolate continuously between discriminative and generative training within a fixed architecture. By ",
  "authors": "Jorge Chang Ortega, Bastien Le Lan, Thomas Serre, Victor Boutin",
  "category": "research",
  "topics": "bias-fairness,safety-alignment,environment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-05-22T16:21:25.000Z",
  "fetched_at": "2026-07-14T16:30:31.923Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/3855",
  "original_url": "https://arxiv.org/abs/2605.23819v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}