{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:IR72YOXH32XCGIIWKMAHZMTZCR","short_pith_number":"pith:IR72YOXH","schema_version":"1.0","canonical_sha256":"447fac3ae7deae23211653007cb279144f2ca395bbcb352b137bcb483124ae00","source":{"kind":"arxiv","id":"2605.22986","version":1},"attestation_state":"computed","paper":{"title":"Robots That Know What to Ask: Recovering Misaligned Rewards through Targeted Explanations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.HC","cs.LG"],"primary_cat":"cs.RO","authors_text":"Andreea Bobu, Helena Merker, Nick Walker","submitted_at":"2026-05-21T19:34:14Z","abstract_excerpt":"Learning reward functions from demonstrations assumes that demonstrations provide adequate supervision over all features -- or task-relevant aspects of behavior. In practice, demonstrations are often imperfect: humans may under-emphasize certain features due to cognitive load or physical difficulty, or the training regime may fail to sufficiently cover all relevant situations. In either case, important features may be underspecified, leading to ambiguity in the learned reward function and misaligned behavior at deployment. We propose a framework that detects such underspecified features and ac"},"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":"2605.22986","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.RO","submitted_at":"2026-05-21T19:34:14Z","cross_cats_sorted":["cs.AI","cs.HC","cs.LG"],"title_canon_sha256":"e1bbf1056017fe0611250577c4913c9ce17a58df7fc09d575cbe839c28fef99c","abstract_canon_sha256":"b5a5fc6889597550e058590aea3827fcabb4ed1cb0fab852c9a644a6986a0f88"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-25T02:01:33.293698Z","signature_b64":"wbTHS5jpV9+gtELec/S17Cf2I5UPVJlm3kepYbO0Hbccc0gM0w7jhDu3OGptgn+w87Jl72VTp0u7PxjJYTINAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"447fac3ae7deae23211653007cb279144f2ca395bbcb352b137bcb483124ae00","last_reissued_at":"2026-05-25T02:01:33.292980Z","signature_status":"signed_v1","first_computed_at":"2026-05-25T02:01:33.292980Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Robots That Know What to Ask: Recovering Misaligned Rewards through Targeted Explanations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.HC","cs.LG"],"primary_cat":"cs.RO","authors_text":"Andreea Bobu, Helena Merker, Nick Walker","submitted_at":"2026-05-21T19:34:14Z","abstract_excerpt":"Learning reward functions from demonstrations assumes that demonstrations provide adequate supervision over all features -- or task-relevant aspects of behavior. In practice, demonstrations are often imperfect: humans may under-emphasize certain features due to cognitive load or physical difficulty, or the training regime may fail to sufficiently cover all relevant situations. In either case, important features may be underspecified, leading to ambiguity in the learned reward function and misaligned behavior at deployment. We propose a framework that detects such underspecified features and ac"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.22986","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/2605.22986/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":"2605.22986","created_at":"2026-05-25T02:01:33.293095+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.22986v1","created_at":"2026-05-25T02:01:33.293095+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.22986","created_at":"2026-05-25T02:01:33.293095+00:00"},{"alias_kind":"pith_short_12","alias_value":"IR72YOXH32XC","created_at":"2026-05-25T02:01:33.293095+00:00"},{"alias_kind":"pith_short_16","alias_value":"IR72YOXH32XCGIIW","created_at":"2026-05-25T02:01:33.293095+00:00"},{"alias_kind":"pith_short_8","alias_value":"IR72YOXH","created_at":"2026-05-25T02:01:33.293095+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/IR72YOXH32XCGIIWKMAHZMTZCR","json":"https://pith.science/pith/IR72YOXH32XCGIIWKMAHZMTZCR.json","graph_json":"https://pith.science/api/pith-number/IR72YOXH32XCGIIWKMAHZMTZCR/graph.json","events_json":"https://pith.science/api/pith-number/IR72YOXH32XCGIIWKMAHZMTZCR/events.json","paper":"https://pith.science/paper/IR72YOXH"},"agent_actions":{"view_html":"https://pith.science/pith/IR72YOXH32XCGIIWKMAHZMTZCR","download_json":"https://pith.science/pith/IR72YOXH32XCGIIWKMAHZMTZCR.json","view_paper":"https://pith.science/paper/IR72YOXH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.22986&json=true","fetch_graph":"https://pith.science/api/pith-number/IR72YOXH32XCGIIWKMAHZMTZCR/graph.json","fetch_events":"https://pith.science/api/pith-number/IR72YOXH32XCGIIWKMAHZMTZCR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IR72YOXH32XCGIIWKMAHZMTZCR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IR72YOXH32XCGIIWKMAHZMTZCR/action/storage_attestation","attest_author":"https://pith.science/pith/IR72YOXH32XCGIIWKMAHZMTZCR/action/author_attestation","sign_citation":"https://pith.science/pith/IR72YOXH32XCGIIWKMAHZMTZCR/action/citation_signature","submit_replication":"https://pith.science/pith/IR72YOXH32XCGIIWKMAHZMTZCR/action/replication_record"}},"created_at":"2026-05-25T02:01:33.293095+00:00","updated_at":"2026-05-25T02:01:33.293095+00:00"}