{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.12324", "kind": "paper", "url": "https://arxiv.org/abs/2608.12324"}, {"id": "gh:fideai/fmg-bench", "kind": "repo", "url": "https://github.com/FideAI/fmg-bench"}, {"id": "hf:fideai/fmg-bench", "kind": "dataset", "url": "https://huggingface.co/datasets/FideAI/fmg-bench"}], "authors": ["Alex Chao"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.12324", "source_url": "https://arxiv.org/abs/2608.12324"}, "categories": ["Biology & Drug Discovery", "Language & Knowledge", "cs.CY"], "description": {"en": "FMG-Bench evaluates English-language Christian theological triage and pastoral guidance in LLMs. It scores whether model responses match expected behaviors for triage levels: primary doctrine, secondary doctrine, prudential, and pastoral, with safety-related escalation appropriate to each scenario. The corpus includes 120 base scenarios plus 37 perturbation variants."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.12324", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.12324", "title": "FMG-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.12324", "modality": null, "name": "FMG-Bench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "unknown", "origin_record_id": "bm_fmg-bench_a89fdc7a", "origin_source": "arxiv", "recognition_confidence": "0.95", "record_sha256": "bd9177eb1927bf45d2b29e714a7c6f0d3ffe024a25f4d70172ff0620aa834848", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2608.12324"}, "publisher": null, "released": "2026-05-29", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.12324", "source_url": "https://arxiv.org/abs/2608.12324"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-12324", "source": "claire_radar", "source_benchmark_id": "2608.12324", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Biology & Drug Discovery"], "area": "Language & Knowledge", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/FideAI/fmg-bench", "githubScope": "benchmark_repo", "githubStars": 3, "hfDailySubmittedAt": null, "hfDataset": "FideAI/fmg-bench", "hfDatasetDownloads": 57, "hfDatasetLikes": 1, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2608.12324", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetDownloads": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-09-30", "state": "stale"}, "hfDatasetLikes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-09-30", "state": "stale"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "state": "unavailable"}}, "status": "fresh"}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "f053736d5b07605f5f849265e4b768d181e9b5d7d243468d4f37f7021760acdb", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "FMG-Bench evaluates English-language Christian theological triage and pastoral guidance in LLMs. It scores whether model responses match expected behaviors for triage levels: primary doctrine, secondary doctrine, prudential, and pastoral, with safety-related escalation appropriate to each scenario. The corpus includes 120 base scenarios plus 37 perturbation variants.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact named benchmark artifact released in abstract", "benchmark release stated in one sentence", "public artifact URL"], "snippet": "We introduce FMG-Bench, the Faith & Moral Guidance Benchmark, a 120-scenario benchmark for evaluating large language model behavior in English-language Christian theological triage and pastoral guidance contexts."}, "familyId": "bmf_9aef90db69a1", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [{"evidence": [{"basis": "existing-code-link", "linkedUrl": "https://github.com/FideAI/fmg-bench", "url": "https://github.com/FideAI/fmg-bench"}, {"basis": "existing-metric-association", "linkedUrl": "https://github.com/FideAI/fmg-bench", "url": "https://huggingface.co/papers/2608.12324"}, {"basis": "linked-from-report", "linkedUrl": "https://github.com/FideAI/fmg-bench", "url": "https://arxiv.org/abs/2608.12324"}], "scope": "benchmark_repo", "url": "https://github.com/FideAI/fmg-bench", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_fmg-bench_a89fdc7a", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Pharma & Biotech"], "links": {"code": "https://github.com/FideAI/fmg-bench", "data": "https://huggingface.co/datasets/FideAI/fmg-bench", "hfPaper": "https://huggingface.co/papers/2608.12324", "paper": "https://arxiv.org/abs/2608.12324", "pdf": "https://arxiv.org/pdf/2608.12324", "project": null, "report": "https://arxiv.org/abs/2608.12324"}, "metrics": [], "motivation": "People increasingly ask large language models (LLMs) for counsel on questions of faith, doctrine, and pastoral care.", "name": "FMG-Bench", "oneLine": "FMG-Bench evaluates English-language Christian theological triage and pastoral guidance in LLMs. It scores whether model responses match expected behaviors for triage levels: primary doctrine, secondary doctrine, prudential, and pastoral, with safety-related escalation appropriate to each scenario. The corpus includes 120 base scenarios plus 37 perturbation variants.", "paperTitle": "When AI Is Your Pastor: A Benchmark for Theological Triage and Pastoral Guidance in Large Language Models", "primaryDomain": "Biology & Drug Discovery", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.12324", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Runnable", "recognitionConfidence": 0.95, "relation": "introduces", "releasedAt": "2026-05-29", "source": {"authors": ["Alex Chao"], "categories": ["cs.CY", "cs.AI", "cs.CL"], "id": "2608.12324", "title": "When AI Is Your Pastor: A Benchmark for Theological Triage and Pastoral Guidance in Large Language Models", "type": "arxiv", "url": "https://arxiv.org/abs/2608.12324"}, "sourceUpdatedAt": "2026-05-29T15:14:25Z", "topics": ["cs.CY"], "venueAttempts": [], "whyItMatters": "Evaluates a niche but real usage area where models are asked for faith and care advice. Provides a structured scoring contract for safety-critical escalation and robustness to rephrasing, which are not covered by generic QA or safety benchmarks. Useful for developers testing models in contexts where human referral matters."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
