{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ZNXBUIQCQCAQVJHC62KOROYXE5","short_pith_number":"pith:ZNXBUIQC","schema_version":"1.0","canonical_sha256":"cb6e1a220280810aa4e2f694e8bb17277c57c53ea3def532469be6ed66633000","source":{"kind":"arxiv","id":"2502.00365","version":1},"attestation_state":"computed","paper":{"title":"What should an AI assessor optimise for?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Daniel Romero-Alvarado, Fernando Mart\\'inez-Plumed, Jos\\'e Hern\\'andez-Orallo","submitted_at":"2025-02-01T08:41:57Z","abstract_excerpt":"An AI assessor is an external, ideally indepen-dent system that predicts an indicator, e.g., a loss value, of another AI system. Assessors can lever-age information from the test results of many other AI systems and have the flexibility of be-ing trained on any loss function or scoring rule: from squared error to toxicity metrics. Here we address the question: is it always optimal to train the assessor for the target metric? Or could it be better to train for a different metric and then map predictions back to the target metric? Us-ing twenty regression and classification problems with tabular"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2502.00365","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-01T08:41:57Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"48b0357b8e0f6cc941a98e83a4fab8ea34008a8e2e0ca390aada5de24ee2c0aa","abstract_canon_sha256":"eede433b77385e32f6b80b273fcaaf73468828af65100b3ca213f10136db0b69"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:08:32.739308Z","signature_b64":"4ORafb83oUadd28Rmx/VkRjLLYYVajaXgV46FekSNFpSh1UgZYU0UGKsEpGpA6XZ5FXlw0oOLQd4qQhm6mCDDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cb6e1a220280810aa4e2f694e8bb17277c57c53ea3def532469be6ed66633000","last_reissued_at":"2026-07-05T10:08:32.738909Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:08:32.738909Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What should an AI assessor optimise for?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Daniel Romero-Alvarado, Fernando Mart\\'inez-Plumed, Jos\\'e Hern\\'andez-Orallo","submitted_at":"2025-02-01T08:41:57Z","abstract_excerpt":"An AI assessor is an external, ideally indepen-dent system that predicts an indicator, e.g., a loss value, of another AI system. Assessors can lever-age information from the test results of many other AI systems and have the flexibility of be-ing trained on any loss function or scoring rule: from squared error to toxicity metrics. Here we address the question: is it always optimal to train the assessor for the target metric? Or could it be better to train for a different metric and then map predictions back to the target metric? Us-ing twenty regression and classification problems with tabular"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.00365","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2502.00365/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2502.00365","created_at":"2026-07-05T10:08:32.738964+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.00365v1","created_at":"2026-07-05T10:08:32.738964+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.00365","created_at":"2026-07-05T10:08:32.738964+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZNXBUIQCQCAQ","created_at":"2026-07-05T10:08:32.738964+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZNXBUIQCQCAQVJHC","created_at":"2026-07-05T10:08:32.738964+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZNXBUIQC","created_at":"2026-07-05T10:08:32.738964+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZNXBUIQCQCAQVJHC62KOROYXE5","json":"https://pith.science/pith/ZNXBUIQCQCAQVJHC62KOROYXE5.json","graph_json":"https://pith.science/api/pith-number/ZNXBUIQCQCAQVJHC62KOROYXE5/graph.json","events_json":"https://pith.science/api/pith-number/ZNXBUIQCQCAQVJHC62KOROYXE5/events.json","paper":"https://pith.science/paper/ZNXBUIQC"},"agent_actions":{"view_html":"https://pith.science/pith/ZNXBUIQCQCAQVJHC62KOROYXE5","download_json":"https://pith.science/pith/ZNXBUIQCQCAQVJHC62KOROYXE5.json","view_paper":"https://pith.science/paper/ZNXBUIQC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.00365&json=true","fetch_graph":"https://pith.science/api/pith-number/ZNXBUIQCQCAQVJHC62KOROYXE5/graph.json","fetch_events":"https://pith.science/api/pith-number/ZNXBUIQCQCAQVJHC62KOROYXE5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZNXBUIQCQCAQVJHC62KOROYXE5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZNXBUIQCQCAQVJHC62KOROYXE5/action/storage_attestation","attest_author":"https://pith.science/pith/ZNXBUIQCQCAQVJHC62KOROYXE5/action/author_attestation","sign_citation":"https://pith.science/pith/ZNXBUIQCQCAQVJHC62KOROYXE5/action/citation_signature","submit_replication":"https://pith.science/pith/ZNXBUIQCQCAQVJHC62KOROYXE5/action/replication_record"}},"created_at":"2026-07-05T10:08:32.738964+00:00","updated_at":"2026-07-05T10:08:32.738964+00:00"}