{
  "abstract": "Text, images, and other rich covariates are increasingly compressed into AI-learned representations and then used as controls in causal analysis. We study when this approach is valid and develop a practical framework for causal inference with learned representations. For a broad class of estimands, an imperfect representation distorts the target causal parameter by the product of two representation errors: one in the outcome regression and one in the balancing weight (or Riesz representer). This yields three constructive results. First, cross-fitted double machine learning (DML) provides valid Wald inference for the representation-dependent target. When representation errors are small, the same interval covers the causal parameter, and it can even attain the semiparametric efficiency bound. Second, fold-wise representation learning (or fine-tuning) is compatible with DML inference for the causal parameter. To this end, we develop convex- and star-aggregation pipelines for learning and combining representations. Third, when representation errors are substantial, we can provide interpretable sensitivity regions and root-$n$ inference for their endpoints. In a multi-modal demand application, seven representation-specific estimates and their star aggregate all imply a negative near-unit elasticity for rank-based price response, and the result remains robust over the reported sensitivity grid.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2610.01935",
  "author_total": 8,
  "authors": [
    {
      "name": "Andres Aradillas Fernandez"
    },
    {
      "name": "Victor Chernozhukov"
    },
    {
      "name": "Carlos Cinelli"
    },
    {
      "name": "Sven Klaassen"
    },
    {
      "name": "Whitney Newey"
    },
    {
      "name": "Martin Spindler"
    },
    {
      "name": "Jan Teichert-Kluge"
    },
    {
      "name": "Suhas Vijaykumar"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/43rkq14wm0m8d.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2610.01935",
  "fulltext_url": null,
  "id": "43rkq14wm0m8d",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2610.01935",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "http://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "econ.EM"
    ],
    [
      "arxiv",
      "stat.ML"
    ],
    [
      "openalex",
      "2613|Advanced Causal Inference Techniques"
    ]
  ],
  "openalex": "W7215683850",
  "pdf_url": "https://arxiv.org/pdf/2610.01935",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-02",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/43rkq14wm0m8d.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:26:56Z",
      "native_id": "2610.01935",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2610.01935"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.ML",
      "subfield": "statistics"
    }
  ],
  "title": "Pragmatic DML with AI-Learned Representations",
  "title_text": "Pragmatic DML with AI-Learned Representations",
  "type": "preprint",
  "updated_at": "2026-10-30T12:26:56Z",
  "url": "https://research.draconitic.com/paper/43rkq14wm0m8d",
  "venue": "arXiv",
  "versions": []
}