{"record": {"aliases": ["FairLens: Benchmarking Fairness in Vision-Language Models for High-Stakes Decision-Making"], "artifacts": [{"id": "arxiv:2609.01691", "kind": "paper", "url": "https://arxiv.org/abs/2609.01691"}, {"kind": "website", "url": "https://vectorinstitute.github.io/fairlens"}], "categories": ["General AI", "Multimodal", "Safety & Trustworthiness"], "description": {"en": "A benchmark and evaluation framework measuring fairness and validity of VLM responses in hiring, legal, and healthcare using over 100K image-question pairs per model across demographic groups."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2609.01691", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2609.01691", "title": "FAIRLENS"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2609.01691", "modality": null, "name": "FAIRLENS", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "false", "origin_record_id": "bm_fairlens-benchmarking-fairness-in-vision-l_bcc6b156", "origin_source": "arxiv", "recognition_confidence": "0.55", "record_sha256": "28870ffdd2b337eefcf6397242feb9b0f758531be3999311452b5302a0a6d52b", "relation": "evaluates_only", "review_model": "deepseek-v4-pro", "review_state": "ai-name-audit-deferred", "reviewed_at": "2026-09-03T06:06:59.591745Z", "source_url": "https://arxiv.org/abs/2609.01691"}, "publisher": null, "released": "2026-09-01", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2609.01691", "source_url": "https://arxiv.org/abs/2609.01691"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2609-01691", "source": "claire_radar", "source_benchmark_id": "2609.01691", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": ["FairLens: Benchmarking Fairness in Vision-Language Models for High-Stakes Decision-Making"], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Safety & Trustworthiness", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/VectorInstitute/fairlens", "githubScope": "benchmark_repo", "githubStars": 0, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 0, "hfPaperUrl": "https://huggingface.co/papers/2609.01691", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-10-01", "state": "stale"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "Fairness evaluation in VLMs for high-stakes domains is a relevant topic, and the project page may attract researchers focused on ethics.", "score": 70}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-03T06:06:59.591745Z", "inputHash": "8c3e01a5b4ede3f875fe5b92d002e105600d987756a536e11e1d5f8eb40e74bf", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "We introduce FAIRLENS, a benchmark and evaluation framework for measuring both the fairness and the validity of VLM responses in three high-stakes domains: hiring, legal, and healthcare.", "canonicalNameSource": "abstract", "decisionReason": "The paper formally names FAIRLENS and provides a project page, but the lack of a fixed scoring contract limits its comparability; still, the benchmark is reusable.", "model": "deepseek-v4-pro", "reviewedAt": "2026-09-03T06:06:59.591745Z", "state": "ai-name-audit-deferred"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "A benchmark and evaluation framework measuring fairness and validity of VLM responses in hiring, legal, and healthcare using over 100K image-question pairs per model across demographic groups.", "displayEligible": false, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["benchmark term in abstract", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL", "named entity is a model, method, framework, or dataset used in evaluation"], "snippet": "We introduce FAIRLENS, a benchmark and evaluation framework for measuring both the fairness and the validity of VLM responses in three high-stakes domains: hiring, legal, and healthcare."}, "familyId": "bmf_026ca207a2f5", "firstSeenAt": "2026-09-03", "heat": null, "id": "bm_fairlens-benchmarking-fairness-in-vision-l_bcc6b156", "indexedAt": "2026-09-03T00:55:28.123857Z", "industrySectors": [], "links": {"code": null, "data": null, "paper": "https://arxiv.org/abs/2609.01691", "pdf": "https://arxiv.org/pdf/2609.01691", "project": "https://vectorinstitute.github.io/fairlens", "report": "https://arxiv.org/abs/2609.01691"}, "metrics": [], "motivation": "Vision-language models (VLMs) are increasingly used to make decisions from visual inputs.", "name": "FAIRLENS", "oneLine": "A benchmark and evaluation framework measuring fairness and validity of VLM responses in hiring, legal, and healthcare using over 100K image-question pairs per model across demographic groups.", "paperTitle": "FairLens: Benchmarking Fairness in Vision-Language Models for High-Stakes Decision-Making", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2609.01691", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-09-03T00:55:28.123857Z"}, "publications": [], "publishers": [{"name": "Vector Institute", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://vectorinstitute.github.io/fairlens"}], "ranking": {"90d": {"confidence": "Medium", "coverage": 0.7, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": 0.0, "value": 0}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.0, "value": 0}}, "confidence": "Medium", "coverage": 0.7, "rank": 577, "score": 15}, "method": "attention-ranking-v13", "rank": 577, "score": 15}}, "readiness": "Paper only", "recognitionConfidence": 0.55, "relation": "evaluates_only", "releasedAt": "2026-09-01", "source": {"authors": ["Vahid Reza Khazaie", "Ahmed Y. Radwan", "Shaina Raza"], "categories": ["cs.CV", "cs.LG"], "id": "2609.01691", "title": "FairLens: Benchmarking Fairness in Vision-Language Models for High-Stakes Decision-Making", "type": "arxiv", "url": "https://arxiv.org/abs/2609.01691"}, "sourceUpdatedAt": "2026-09-01T16:06:50Z", "topics": ["Multimodal"], "venueAttempts": [], "whyItMatters": "It highlights that unwarranted inference is a prevalent failure and provides a complementary view of fairness beyond parity metrics, which is crucial for high-stakes applications."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
