{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:Y4HOJNDX2XFUIGUF2L5T6FFQW7","short_pith_number":"pith:Y4HOJNDX","schema_version":"1.0","canonical_sha256":"c70ee4b477d5cb441a85d2fb3f14b0b7ec30870695f3627d28b1e69a52b90154","source":{"kind":"arxiv","id":"2010.12016","version":1},"attestation_state":"computed","paper":{"title":"Towards falsifiable interpretability research","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.LG","stat.ML"],"primary_cat":"cs.CY","authors_text":"Ari Morcos, Matthew L. Leavitt","submitted_at":"2020-10-22T22:03:41Z","abstract_excerpt":"Methods for understanding the decisions of and mechanisms underlying deep neural networks (DNNs) typically rely on building intuition by emphasizing sensory or semantic features of individual examples. For instance, methods aim to visualize the components of an input which are \"important\" to a network's decision, or to measure the semantic properties of single neurons. Here, we argue that interpretability research suffers from an over-reliance on intuition-based approaches that risk-and in some cases have caused-illusory progress and misleading conclusions. We identify a set of limitations tha"},"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":"2010.12016","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CY","submitted_at":"2020-10-22T22:03:41Z","cross_cats_sorted":["cs.AI","cs.CV","cs.LG","stat.ML"],"title_canon_sha256":"0f888cfaec5d178c5429c6469c6959c268f4d7a993238d455de6aa3f4f488ca1","abstract_canon_sha256":"9771acf4dc564db3db2c6c1132a806c5d2ecedfb52243355d225fda17110f581"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:45:32.794415Z","signature_b64":"hCLPXherd5wDxnFAUMJ4cG7xofPl4NWoCdmrXb09xjjmNucFVrCy/k1fQGCLOIedWkRPteFAbX7vnLBKgMOXBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c70ee4b477d5cb441a85d2fb3f14b0b7ec30870695f3627d28b1e69a52b90154","last_reissued_at":"2026-07-05T01:45:32.793875Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:45:32.793875Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards falsifiable interpretability research","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.LG","stat.ML"],"primary_cat":"cs.CY","authors_text":"Ari Morcos, Matthew L. Leavitt","submitted_at":"2020-10-22T22:03:41Z","abstract_excerpt":"Methods for understanding the decisions of and mechanisms underlying deep neural networks (DNNs) typically rely on building intuition by emphasizing sensory or semantic features of individual examples. For instance, methods aim to visualize the components of an input which are \"important\" to a network's decision, or to measure the semantic properties of single neurons. Here, we argue that interpretability research suffers from an over-reliance on intuition-based approaches that risk-and in some cases have caused-illusory progress and misleading conclusions. We identify a set of limitations tha"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2010.12016","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/2010.12016/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":"2010.12016","created_at":"2026-07-05T01:45:32.793944+00:00"},{"alias_kind":"arxiv_version","alias_value":"2010.12016v1","created_at":"2026-07-05T01:45:32.793944+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2010.12016","created_at":"2026-07-05T01:45:32.793944+00:00"},{"alias_kind":"pith_short_12","alias_value":"Y4HOJNDX2XFU","created_at":"2026-07-05T01:45:32.793944+00:00"},{"alias_kind":"pith_short_16","alias_value":"Y4HOJNDX2XFUIGUF","created_at":"2026-07-05T01:45:32.793944+00:00"},{"alias_kind":"pith_short_8","alias_value":"Y4HOJNDX","created_at":"2026-07-05T01:45:32.793944+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.00267","citing_title":"Validating Causal Abstraction Metrics on Simulated Complex Systems","ref_index":197,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Y4HOJNDX2XFUIGUF2L5T6FFQW7","json":"https://pith.science/pith/Y4HOJNDX2XFUIGUF2L5T6FFQW7.json","graph_json":"https://pith.science/api/pith-number/Y4HOJNDX2XFUIGUF2L5T6FFQW7/graph.json","events_json":"https://pith.science/api/pith-number/Y4HOJNDX2XFUIGUF2L5T6FFQW7/events.json","paper":"https://pith.science/paper/Y4HOJNDX"},"agent_actions":{"view_html":"https://pith.science/pith/Y4HOJNDX2XFUIGUF2L5T6FFQW7","download_json":"https://pith.science/pith/Y4HOJNDX2XFUIGUF2L5T6FFQW7.json","view_paper":"https://pith.science/paper/Y4HOJNDX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2010.12016&json=true","fetch_graph":"https://pith.science/api/pith-number/Y4HOJNDX2XFUIGUF2L5T6FFQW7/graph.json","fetch_events":"https://pith.science/api/pith-number/Y4HOJNDX2XFUIGUF2L5T6FFQW7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Y4HOJNDX2XFUIGUF2L5T6FFQW7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Y4HOJNDX2XFUIGUF2L5T6FFQW7/action/storage_attestation","attest_author":"https://pith.science/pith/Y4HOJNDX2XFUIGUF2L5T6FFQW7/action/author_attestation","sign_citation":"https://pith.science/pith/Y4HOJNDX2XFUIGUF2L5T6FFQW7/action/citation_signature","submit_replication":"https://pith.science/pith/Y4HOJNDX2XFUIGUF2L5T6FFQW7/action/replication_record"}},"created_at":"2026-07-05T01:45:32.793944+00:00","updated_at":"2026-07-05T01:45:32.793944+00:00"}