{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:KQ5LNWYLWTLH7BYNKADEEC6T6Y","short_pith_number":"pith:KQ5LNWYL","schema_version":"1.0","canonical_sha256":"543ab6db0bb4d67f870d5006420bd3f61ccbf97fd013250a09f1dd704758a818","source":{"kind":"arxiv","id":"2504.11844","version":1},"attestation_state":"computed","paper":{"title":"Evaluating the Goal-Directedness of Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Alexis Bellot, Cristina Garbacea, Henry Papadatos, Jonathan Richens, Rohin Shah, Sim\\'eon Campos, Tom Everitt","submitted_at":"2025-04-16T08:07:08Z","abstract_excerpt":"To what extent do LLMs use their capabilities towards their given goal? We take this as a measure of their goal-directedness. We evaluate goal-directedness on tasks that require information gathering, cognitive effort, and plan execution, where we use subtasks to infer each model's relevant capabilities. Our evaluations of LLMs from Google DeepMind, OpenAI, and Anthropic show that goal-directedness is relatively consistent across tasks, differs from task performance, and is only moderately sensitive to motivational prompts. Notably, most models are not fully goal-directed. We hope our goal-dir"},"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":"2504.11844","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-04-16T08:07:08Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"5c9ede1b730c8a76681db0da3e10fc9ace3bdc2dceb0eb49b98d0255b99871a6","abstract_canon_sha256":"fa305dfc60818ff54fbdf607a023425d55733ae2ab747ae2ad4760ca56796bf1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:49:58.591655Z","signature_b64":"wNwzznw0F7Z5zMLDSpRuAmy60Cuq+8b3YxIjzrAV7EMM704eQpGwXwO9akBTO8QGD8Tv9dY7VnqAFxDeqhfvCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"543ab6db0bb4d67f870d5006420bd3f61ccbf97fd013250a09f1dd704758a818","last_reissued_at":"2026-07-05T10:49:58.591041Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:49:58.591041Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Evaluating the Goal-Directedness of Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Alexis Bellot, Cristina Garbacea, Henry Papadatos, Jonathan Richens, Rohin Shah, Sim\\'eon Campos, Tom Everitt","submitted_at":"2025-04-16T08:07:08Z","abstract_excerpt":"To what extent do LLMs use their capabilities towards their given goal? We take this as a measure of their goal-directedness. We evaluate goal-directedness on tasks that require information gathering, cognitive effort, and plan execution, where we use subtasks to infer each model's relevant capabilities. Our evaluations of LLMs from Google DeepMind, OpenAI, and Anthropic show that goal-directedness is relatively consistent across tasks, differs from task performance, and is only moderately sensitive to motivational prompts. Notably, most models are not fully goal-directed. We hope our goal-dir"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.11844","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/2504.11844/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":"2504.11844","created_at":"2026-07-05T10:49:58.591116+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.11844v1","created_at":"2026-07-05T10:49:58.591116+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.11844","created_at":"2026-07-05T10:49:58.591116+00:00"},{"alias_kind":"pith_short_12","alias_value":"KQ5LNWYLWTLH","created_at":"2026-07-05T10:49:58.591116+00:00"},{"alias_kind":"pith_short_16","alias_value":"KQ5LNWYLWTLH7BYN","created_at":"2026-07-05T10:49:58.591116+00:00"},{"alias_kind":"pith_short_8","alias_value":"KQ5LNWYL","created_at":"2026-07-05T10:49:58.591116+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.00248","citing_title":"Causal Foundations of Collective Agency","ref_index":18,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KQ5LNWYLWTLH7BYNKADEEC6T6Y","json":"https://pith.science/pith/KQ5LNWYLWTLH7BYNKADEEC6T6Y.json","graph_json":"https://pith.science/api/pith-number/KQ5LNWYLWTLH7BYNKADEEC6T6Y/graph.json","events_json":"https://pith.science/api/pith-number/KQ5LNWYLWTLH7BYNKADEEC6T6Y/events.json","paper":"https://pith.science/paper/KQ5LNWYL"},"agent_actions":{"view_html":"https://pith.science/pith/KQ5LNWYLWTLH7BYNKADEEC6T6Y","download_json":"https://pith.science/pith/KQ5LNWYLWTLH7BYNKADEEC6T6Y.json","view_paper":"https://pith.science/paper/KQ5LNWYL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.11844&json=true","fetch_graph":"https://pith.science/api/pith-number/KQ5LNWYLWTLH7BYNKADEEC6T6Y/graph.json","fetch_events":"https://pith.science/api/pith-number/KQ5LNWYLWTLH7BYNKADEEC6T6Y/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KQ5LNWYLWTLH7BYNKADEEC6T6Y/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KQ5LNWYLWTLH7BYNKADEEC6T6Y/action/storage_attestation","attest_author":"https://pith.science/pith/KQ5LNWYLWTLH7BYNKADEEC6T6Y/action/author_attestation","sign_citation":"https://pith.science/pith/KQ5LNWYLWTLH7BYNKADEEC6T6Y/action/citation_signature","submit_replication":"https://pith.science/pith/KQ5LNWYLWTLH7BYNKADEEC6T6Y/action/replication_record"}},"created_at":"2026-07-05T10:49:58.591116+00:00","updated_at":"2026-07-05T10:49:58.591116+00:00"}