{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2016:OTCIM2WH27COCLBA6EIB7FGOYE","short_pith_number":"pith:OTCIM2WH","schema_version":"1.0","canonical_sha256":"74c4866ac7d7c4e12c20f1101f94cec11e43f31998ae3fa95963fdaab4325750","source":{"kind":"arxiv","id":"1605.03143","version":1},"attestation_state":"computed","paper":{"title":"Avoiding Wireheading with Value Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Marcus Hutter, Tom Everitt","submitted_at":"2016-05-10T18:28:57Z","abstract_excerpt":"How can we design good goals for arbitrarily intelligent agents? Reinforcement learning (RL) is a natural approach. Unfortunately, RL does not work well for generally intelligent agents, as RL agents are incentivised to shortcut the reward sensor for maximum reward -- the so-called wireheading problem. In this paper we suggest an alternative to RL called value reinforcement learning (VRL). In VRL, agents use the reward signal to learn a utility function. The VRL setup allows us to remove the incentive to wirehead by placing a constraint on the agent's actions. The constraint is defined in term"},"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":"1605.03143","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2016-05-10T18:28:57Z","cross_cats_sorted":[],"title_canon_sha256":"e6cc9ead74d3467ccac97be7e771faf8fd831a05966671e53ba63df7aad24728","abstract_canon_sha256":"0f299bacaf603ec01594c49da7691b7d61c34222f66265e451ac6eea3d2f9086"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T01:15:07.436666Z","signature_b64":"549GyB07SlzBeo/Wm1KE3bNKCxNSFIgatmOxLCu6v2PuBNpI6vfq6TlysNR+N7s4xNfkmRoxzNtIDsJJNyGaAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"74c4866ac7d7c4e12c20f1101f94cec11e43f31998ae3fa95963fdaab4325750","last_reissued_at":"2026-05-18T01:15:07.435920Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T01:15:07.435920Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Avoiding Wireheading with Value Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Marcus Hutter, Tom Everitt","submitted_at":"2016-05-10T18:28:57Z","abstract_excerpt":"How can we design good goals for arbitrarily intelligent agents? Reinforcement learning (RL) is a natural approach. Unfortunately, RL does not work well for generally intelligent agents, as RL agents are incentivised to shortcut the reward sensor for maximum reward -- the so-called wireheading problem. In this paper we suggest an alternative to RL called value reinforcement learning (VRL). In VRL, agents use the reward signal to learn a utility function. The VRL setup allows us to remove the incentive to wirehead by placing a constraint on the agent's actions. The constraint is defined in term"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1605.03143","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":""},"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":"1605.03143","created_at":"2026-05-18T01:15:07.436040+00:00"},{"alias_kind":"arxiv_version","alias_value":"1605.03143v1","created_at":"2026-05-18T01:15:07.436040+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1605.03143","created_at":"2026-05-18T01:15:07.436040+00:00"},{"alias_kind":"pith_short_12","alias_value":"OTCIM2WH27CO","created_at":"2026-05-18T12:30:36.002864+00:00"},{"alias_kind":"pith_short_16","alias_value":"OTCIM2WH27COCLBA","created_at":"2026-05-18T12:30:36.002864+00:00"},{"alias_kind":"pith_short_8","alias_value":"OTCIM2WH","created_at":"2026-05-18T12:30:36.002864+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"1606.06565","citing_title":"Concrete Problems in AI Safety","ref_index":49,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OTCIM2WH27COCLBA6EIB7FGOYE","json":"https://pith.science/pith/OTCIM2WH27COCLBA6EIB7FGOYE.json","graph_json":"https://pith.science/api/pith-number/OTCIM2WH27COCLBA6EIB7FGOYE/graph.json","events_json":"https://pith.science/api/pith-number/OTCIM2WH27COCLBA6EIB7FGOYE/events.json","paper":"https://pith.science/paper/OTCIM2WH"},"agent_actions":{"view_html":"https://pith.science/pith/OTCIM2WH27COCLBA6EIB7FGOYE","download_json":"https://pith.science/pith/OTCIM2WH27COCLBA6EIB7FGOYE.json","view_paper":"https://pith.science/paper/OTCIM2WH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1605.03143&json=true","fetch_graph":"https://pith.science/api/pith-number/OTCIM2WH27COCLBA6EIB7FGOYE/graph.json","fetch_events":"https://pith.science/api/pith-number/OTCIM2WH27COCLBA6EIB7FGOYE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OTCIM2WH27COCLBA6EIB7FGOYE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OTCIM2WH27COCLBA6EIB7FGOYE/action/storage_attestation","attest_author":"https://pith.science/pith/OTCIM2WH27COCLBA6EIB7FGOYE/action/author_attestation","sign_citation":"https://pith.science/pith/OTCIM2WH27COCLBA6EIB7FGOYE/action/citation_signature","submit_replication":"https://pith.science/pith/OTCIM2WH27COCLBA6EIB7FGOYE/action/replication_record"}},"created_at":"2026-05-18T01:15:07.436040+00:00","updated_at":"2026-05-18T01:15:07.436040+00:00"}