{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2021:6E62MATHSAQ57GTS4XG27ATHYT","short_pith_number":"pith:6E62MATH","canonical_record":{"source":{"id":"2108.07258","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-08-16T17:50:08Z","cross_cats_sorted":["cs.AI","cs.CY"],"title_canon_sha256":"a84869a820663c46ddb29aea8dc7d0ddc5a681c10bfc42cbd46f04543af60a8e","abstract_canon_sha256":"581a55584132a515bbe8150b92d0533a6df7fe31f5193d164cc6d189726181a1"},"schema_version":"1.0"},"canonical_sha256":"f13da602679021df9a72e5cdaf8267c4ebcd8475e8983c4ac498067a7250b8aa","source":{"kind":"arxiv","id":"2108.07258","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2108.07258","created_at":"2026-07-05T04:39:47Z"},{"alias_kind":"arxiv_version","alias_value":"2108.07258v3","created_at":"2026-07-05T04:39:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2108.07258","created_at":"2026-07-05T04:39:47Z"},{"alias_kind":"pith_short_12","alias_value":"6E62MATHSAQ5","created_at":"2026-07-05T04:39:47Z"},{"alias_kind":"pith_short_16","alias_value":"6E62MATHSAQ57GTS","created_at":"2026-07-05T04:39:47Z"},{"alias_kind":"pith_short_8","alias_value":"6E62MATH","created_at":"2026-07-05T04:39:47Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2021:6E62MATHSAQ57GTS4XG27ATHYT","target":"record","payload":{"canonical_record":{"source":{"id":"2108.07258","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-08-16T17:50:08Z","cross_cats_sorted":["cs.AI","cs.CY"],"title_canon_sha256":"a84869a820663c46ddb29aea8dc7d0ddc5a681c10bfc42cbd46f04543af60a8e","abstract_canon_sha256":"581a55584132a515bbe8150b92d0533a6df7fe31f5193d164cc6d189726181a1"},"schema_version":"1.0"},"canonical_sha256":"f13da602679021df9a72e5cdaf8267c4ebcd8475e8983c4ac498067a7250b8aa","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:39:47.644678Z","signature_b64":"XVHpsSikfRopduALOMlbx//2aulDhN9hJN2zuYHdYm/VcXSd59LKJYjb112pmqsJHvxATdfjRMrHnu71q2EoDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f13da602679021df9a72e5cdaf8267c4ebcd8475e8983c4ac498067a7250b8aa","last_reissued_at":"2026-07-05T04:39:47.644175Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:39:47.644175Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2108.07258","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T04:39:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Mbb84CECyASGLieNOTgTJq0x9oYQDio8brrwbk1VTxN2aWV96SkkWIw8HjX2x8ATeabQ1uhSfeKdmLp/sFHiAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-03T22:51:51.324236Z"},"content_sha256":"4f2e872486a39abf0814323b856a40ccbf16c5389fefbf9e5a96231ad47a4fc8","schema_version":"1.0","event_id":"sha256:4f2e872486a39abf0814323b856a40ccbf16c5389fefbf9e5a96231ad47a4fc8"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2021:6E62MATHSAQ57GTS4XG27ATHYT","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"On the Opportunities and Risks of Foundation Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Foundation models trained on broad data at scale develop emergent capabilities that incentivize their use across many tasks while passing any defects to all downstream adaptations.","cross_cats":["cs.AI","cs.CY"],"primary_cat":"cs.LG","authors_text":"Aditi Raghunathan, Alex Tamkin, Ali Malik, Allen Nie, Ananya Kumar, Andy Shih, Annie Chen, Antoine Bosselut, Armin W. Thomas, Avanika Narayan, Ben Newman, Bohan Wu, Camilo Ruiz, Chelsea Finn, Chris Donahue, Chris Piech, Christopher D. Manning, Christopher Potts, Christopher R\\'e, Dallas Card, Daniel E. Ho, Dan Jurafsky, Deepak Narayanan, Dora Demszky, Dorsa Sadigh, Drew A. Hudson, Ehsan Adeli, Emma Brunskill, Eric Mitchell, Erik Brynjolfsson, Esin Durmus, Eva Portelance, Faisal Ladhak, Fereshte Khani, Florian Tram\\`er, Frieda Rong, Geoff Keeling, Giray Ogut, Hamed Nilforoshan, Hongyu Ren, Isabelle Levent, Isabel Papadimitriou, Jack Ryan, Jared Quincy Davis, Jeannette Bohg, Jenny Hong, Jiajun Wu, Jiaxuan You, Jing Huang, John Etchemendy, John Hewitt, Joon Sung Park, Juan Carlos Niebles, Julian Nyarko, Jure Leskovec, Kaitlyn Zhou, Karan Goel, Kathleen Creel, Kawin Ethayarajh, Keshav Santhanam, Krishnan Srinivasan, Kyle Hsu, Laurel Orr, Lauren Gillespie, Li Fei-Fei, Lucia Zheng, Mark Krass, Matei Zaharia, Michael S. Bernstein, Michael Zhang, Michihiro Yasunaga, Mina Lee, Moussa Doumbouya, Neel Guha, Niladri Chatterji, Noah Goodman, Omar Khattab, Pang Wei Koh, Percy Liang, Peter Henderson, Pratyusha Kalluri, Ranjay Krishna, Rishi Bommasani, Rob Reich, Rodrigo Castellon, Rohan Taori, Rohith Kuditipudi, Rose E. Wang, Russ Altman, Saahil Jain, Sang Michael Xie, Shelby Grossman, Shiori Sagawa, Shyamal Buch, Siddharth Karamcheti, Simran Arora, Stefano Ermon, Suraj Nair, Suvir Mirchandani, Sydney von Arx, Tatsunori Hashimoto, Tengyu Ma, Thomas Icard, Tianyi Zhang, Tony Lee, Trevor Gale, William Wang, Xiang Lisa Li, Xikun Zhang, Xuechen Li, Yuhuai Wu, Yuhui Zhang, Yusuf Roohani, Zanele Munyikwa","submitted_at":"2021-08-16T17:50:08Z","abstract_excerpt":"AI is undergoing a paradigm shift with the rise of models (e.g., BERT, DALL-E, GPT-3) that are trained on broad data at scale and are adaptable to a wide range of downstream tasks. We call these models foundation models to underscore their critically central yet incomplete character. This report provides a thorough account of the opportunities and risks of foundation models, ranging from their capabilities (e.g., language, vision, robotics, reasoning, human interaction) and technical principles(e.g., model architectures, training procedures, data, systems, security, evaluation, theory) to thei"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Though foundation models are based on standard deep learning and transfer learning, their scale results in new emergent capabilities, and their effectiveness across so many tasks incentivizes homogenization. Homogenization provides powerful leverage but demands caution, as the defects of the foundation model are inherited by all the adapted models downstream.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the scale of training produces genuinely emergent capabilities that are not predictable from smaller models and that widespread adoption will lead to homogenization without adequate independent safeguards or understanding.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Foundation models are large adaptable AI systems with emergent capabilities that offer broad opportunities but carry risks from homogenization, opacity, and inherited defects across downstream applications.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Foundation models trained on broad data at scale develop emergent capabilities that incentivize their use across many tasks while passing any defects to all downstream adaptations.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"33c653d150d59c3a9fd9bd812aea351080e339fdc52aa1884c76a156465d215e"},"source":{"id":"2108.07258","kind":"arxiv","version":3},"verdict":{"id":"d15c928c-9f8f-4531-896c-796ddae174ba","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T16:10:14.335079Z","strongest_claim":"Though foundation models are based on standard deep learning and transfer learning, their scale results in new emergent capabilities, and their effectiveness across so many tasks incentivizes homogenization. Homogenization provides powerful leverage but demands caution, as the defects of the foundation model are inherited by all the adapted models downstream.","one_line_summary":"Foundation models are large adaptable AI systems with emergent capabilities that offer broad opportunities but carry risks from homogenization, opacity, and inherited defects across downstream applications.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the scale of training produces genuinely emergent capabilities that are not predictable from smaller models and that widespread adoption will lead to homogenization without adequate independent safeguards or understanding.","pith_extraction_headline":"Foundation models trained on broad data at scale develop emergent capabilities that incentivize their use across many tasks while passing any defects to all downstream adaptations."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2108.07258/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":18,"sample":[{"doi":"","year":2021,"title":"Invariant Risk Minimization","work_id":"d76c6842-b84d-44ec-bcea-b80cd8d07981","ref_index":1,"cited_arxiv_id":"1907.02893","is_internal_anchor":true},{"doi":"","year":2015,"title":"Neural Machine Translation by Jointly Learning to Align and Translate","work_id":"d831e763-d530-4029-a65c-ac595d82cb2a","ref_index":2,"cited_arxiv_id":"1409.0473","is_internal_anchor":true},{"doi":"10.1007/s10993-012-9245-8","year":1986,"title":"Documentation Debt","work_id":"73bb7cd4-8a43-448e-895a-975640a41dd9","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.18653/v1/2020.acl-main.431","year":2017,"title":"cost disease","work_id":"a6246246-f602-49cb-a4f3-1f365ccddaed","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"Chunyuan Deng, Yilun Zhao, Xiangru Tang, Mark Gerstein, and Arman Cohan","work_id":"2662531a-bd8c-4a98-9104-354aba180472","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":18,"snapshot_sha256":"b09a1676f14d146ff19a1292f3a75b2384446e8aea12a1e82973ac631b895ac8","internal_anchors":2},"formal_canon":{"evidence_count":1,"snapshot_sha256":"235b00634bbb241faa0ae4f0c37f26ec38f651648ad5ff206f5508d46a89de73"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"d15c928c-9f8f-4531-896c-796ddae174ba"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T04:39:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"KVtCBDrUB+GFvFB9buG1iirfuM4BzLEcUn1lHCaROd88mk8nMN6fums0NaH2IyNaL4mj+S44mHAFJTZbGBi1Bw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-03T22:51:51.325137Z"},"content_sha256":"33d98ba80a61f01372f4cc083676a148b9a2a558ed27c93998dc7e979a89795a","schema_version":"1.0","event_id":"sha256:33d98ba80a61f01372f4cc083676a148b9a2a558ed27c93998dc7e979a89795a"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/6E62MATHSAQ57GTS4XG27ATHYT/bundle.json","state_url":"https://pith.science/pith/6E62MATHSAQ57GTS4XG27ATHYT/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/6E62MATHSAQ57GTS4XG27ATHYT/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-03T22:51:51Z","links":{"resolver":"https://pith.science/pith/6E62MATHSAQ57GTS4XG27ATHYT","bundle":"https://pith.science/pith/6E62MATHSAQ57GTS4XG27ATHYT/bundle.json","state":"https://pith.science/pith/6E62MATHSAQ57GTS4XG27ATHYT/state.json","well_known_bundle":"https://pith.science/.well-known/pith/6E62MATHSAQ57GTS4XG27ATHYT/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2021:6E62MATHSAQ57GTS4XG27ATHYT","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"581a55584132a515bbe8150b92d0533a6df7fe31f5193d164cc6d189726181a1","cross_cats_sorted":["cs.AI","cs.CY"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-08-16T17:50:08Z","title_canon_sha256":"a84869a820663c46ddb29aea8dc7d0ddc5a681c10bfc42cbd46f04543af60a8e"},"schema_version":"1.0","source":{"id":"2108.07258","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2108.07258","created_at":"2026-07-05T04:39:47Z"},{"alias_kind":"arxiv_version","alias_value":"2108.07258v3","created_at":"2026-07-05T04:39:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2108.07258","created_at":"2026-07-05T04:39:47Z"},{"alias_kind":"pith_short_12","alias_value":"6E62MATHSAQ5","created_at":"2026-07-05T04:39:47Z"},{"alias_kind":"pith_short_16","alias_value":"6E62MATHSAQ57GTS","created_at":"2026-07-05T04:39:47Z"},{"alias_kind":"pith_short_8","alias_value":"6E62MATH","created_at":"2026-07-05T04:39:47Z"}],"graph_snapshots":[{"event_id":"sha256:33d98ba80a61f01372f4cc083676a148b9a2a558ed27c93998dc7e979a89795a","target":"graph","created_at":"2026-07-05T04:39:47Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Though foundation models are based on standard deep learning and transfer learning, their scale results in new emergent capabilities, and their effectiveness across so many tasks incentivizes homogenization. Homogenization provides powerful leverage but demands caution, as the defects of the foundation model are inherited by all the adapted models downstream."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the scale of training produces genuinely emergent capabilities that are not predictable from smaller models and that widespread adoption will lead to homogenization without adequate independent safeguards or understanding."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Foundation models are large adaptable AI systems with emergent capabilities that offer broad opportunities but carry risks from homogenization, opacity, and inherited defects across downstream applications."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Foundation models trained on broad data at scale develop emergent capabilities that incentivize their use across many tasks while passing any defects to all downstream adaptations."}],"snapshot_sha256":"33c653d150d59c3a9fd9bd812aea351080e339fdc52aa1884c76a156465d215e"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"235b00634bbb241faa0ae4f0c37f26ec38f651648ad5ff206f5508d46a89de73"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2108.07258/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"AI is undergoing a paradigm shift with the rise of models (e.g., BERT, DALL-E, GPT-3) that are trained on broad data at scale and are adaptable to a wide range of downstream tasks. We call these models foundation models to underscore their critically central yet incomplete character. This report provides a thorough account of the opportunities and risks of foundation models, ranging from their capabilities (e.g., language, vision, robotics, reasoning, human interaction) and technical principles(e.g., model architectures, training procedures, data, systems, security, evaluation, theory) to thei","authors_text":"Aditi Raghunathan, Alex Tamkin, Ali Malik, Allen Nie, Ananya Kumar, Andy Shih, Annie Chen, Antoine Bosselut, Armin W. Thomas, Avanika Narayan, Ben Newman, Bohan Wu, Camilo Ruiz, Chelsea Finn, Chris Donahue, Chris Piech, Christopher D. Manning, Christopher Potts, Christopher R\\'e, Dallas Card, Daniel E. Ho, Dan Jurafsky, Deepak Narayanan, Dora Demszky, Dorsa Sadigh, Drew A. Hudson, Ehsan Adeli, Emma Brunskill, Eric Mitchell, Erik Brynjolfsson, Esin Durmus, Eva Portelance, Faisal Ladhak, Fereshte Khani, Florian Tram\\`er, Frieda Rong, Geoff Keeling, Giray Ogut, Hamed Nilforoshan, Hongyu Ren, Isabelle Levent, Isabel Papadimitriou, Jack Ryan, Jared Quincy Davis, Jeannette Bohg, Jenny Hong, Jiajun Wu, Jiaxuan You, Jing Huang, John Etchemendy, John Hewitt, Joon Sung Park, Juan Carlos Niebles, Julian Nyarko, Jure Leskovec, Kaitlyn Zhou, Karan Goel, Kathleen Creel, Kawin Ethayarajh, Keshav Santhanam, Krishnan Srinivasan, Kyle Hsu, Laurel Orr, Lauren Gillespie, Li Fei-Fei, Lucia Zheng, Mark Krass, Matei Zaharia, Michael S. Bernstein, Michael Zhang, Michihiro Yasunaga, Mina Lee, Moussa Doumbouya, Neel Guha, Niladri Chatterji, Noah Goodman, Omar Khattab, Pang Wei Koh, Percy Liang, Peter Henderson, Pratyusha Kalluri, Ranjay Krishna, Rishi Bommasani, Rob Reich, Rodrigo Castellon, Rohan Taori, Rohith Kuditipudi, Rose E. Wang, Russ Altman, Saahil Jain, Sang Michael Xie, Shelby Grossman, Shiori Sagawa, Shyamal Buch, Siddharth Karamcheti, Simran Arora, Stefano Ermon, Suraj Nair, Suvir Mirchandani, Sydney von Arx, Tatsunori Hashimoto, Tengyu Ma, Thomas Icard, Tianyi Zhang, Tony Lee, Trevor Gale, William Wang, Xiang Lisa Li, Xikun Zhang, Xuechen Li, Yuhuai Wu, Yuhui Zhang, Yusuf Roohani, Zanele Munyikwa","cross_cats":["cs.AI","cs.CY"],"headline":"Foundation models trained on broad data at scale develop emergent capabilities that incentivize their use across many tasks while passing any defects to all downstream adaptations.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-08-16T17:50:08Z","title":"On the Opportunities and Risks of Foundation Models"},"references":{"count":18,"internal_anchors":2,"resolved_work":18,"sample":[{"cited_arxiv_id":"1907.02893","doi":"","is_internal_anchor":true,"ref_index":1,"title":"Invariant Risk Minimization","work_id":"d76c6842-b84d-44ec-bcea-b80cd8d07981","year":2021},{"cited_arxiv_id":"1409.0473","doi":"","is_internal_anchor":true,"ref_index":2,"title":"Neural Machine Translation by Jointly Learning to Align and Translate","work_id":"d831e763-d530-4029-a65c-ac595d82cb2a","year":2015},{"cited_arxiv_id":"","doi":"10.1007/s10993-012-9245-8","is_internal_anchor":false,"ref_index":3,"title":"Documentation Debt","work_id":"73bb7cd4-8a43-448e-895a-975640a41dd9","year":1986},{"cited_arxiv_id":"","doi":"10.18653/v1/2020.acl-main.431","is_internal_anchor":false,"ref_index":4,"title":"cost disease","work_id":"a6246246-f602-49cb-a4f3-1f365ccddaed","year":2017},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Chunyuan Deng, Yilun Zhao, Xiangru Tang, Mark Gerstein, and Arman Cohan","work_id":"2662531a-bd8c-4a98-9104-354aba180472","year":2021}],"snapshot_sha256":"b09a1676f14d146ff19a1292f3a75b2384446e8aea12a1e82973ac631b895ac8"},"source":{"id":"2108.07258","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-10T16:10:14.335079Z","id":"d15c928c-9f8f-4531-896c-796ddae174ba","model_set":{"reader":"grok-4.3"},"one_line_summary":"Foundation models are large adaptable AI systems with emergent capabilities that offer broad opportunities but carry risks from homogenization, opacity, and inherited defects across downstream applications.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Foundation models trained on broad data at scale develop emergent capabilities that incentivize their use across many tasks while passing any defects to all downstream adaptations.","strongest_claim":"Though foundation models are based on standard deep learning and transfer learning, their scale results in new emergent capabilities, and their effectiveness across so many tasks incentivizes homogenization. Homogenization provides powerful leverage but demands caution, as the defects of the foundation model are inherited by all the adapted models downstream.","weakest_assumption":"That the scale of training produces genuinely emergent capabilities that are not predictable from smaller models and that widespread adoption will lead to homogenization without adequate independent safeguards or understanding."}},"verdict_id":"d15c928c-9f8f-4531-896c-796ddae174ba"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:4f2e872486a39abf0814323b856a40ccbf16c5389fefbf9e5a96231ad47a4fc8","target":"record","created_at":"2026-07-05T04:39:47Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"581a55584132a515bbe8150b92d0533a6df7fe31f5193d164cc6d189726181a1","cross_cats_sorted":["cs.AI","cs.CY"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-08-16T17:50:08Z","title_canon_sha256":"a84869a820663c46ddb29aea8dc7d0ddc5a681c10bfc42cbd46f04543af60a8e"},"schema_version":"1.0","source":{"id":"2108.07258","kind":"arxiv","version":3}},"canonical_sha256":"f13da602679021df9a72e5cdaf8267c4ebcd8475e8983c4ac498067a7250b8aa","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f13da602679021df9a72e5cdaf8267c4ebcd8475e8983c4ac498067a7250b8aa","first_computed_at":"2026-07-05T04:39:47.644175Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T04:39:47.644175Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"XVHpsSikfRopduALOMlbx//2aulDhN9hJN2zuYHdYm/VcXSd59LKJYjb112pmqsJHvxATdfjRMrHnu71q2EoDw==","signature_status":"signed_v1","signed_at":"2026-07-05T04:39:47.644678Z","signed_message":"canonical_sha256_bytes"},"source_id":"2108.07258","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4f2e872486a39abf0814323b856a40ccbf16c5389fefbf9e5a96231ad47a4fc8","sha256:33d98ba80a61f01372f4cc083676a148b9a2a558ed27c93998dc7e979a89795a"],"state_sha256":"7400b9b5174a6b61c079217cbf98c8a742c439657eb8c1878c44efd50fba91e0"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"qPshvj7n+rtxgqIb2NnXuGyyj1X1ZNzk/Vb8guaxZkqSrDwBP9fzAAn/AxCHhlihBrzn5aCqp1bz7JAzqTL9AA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-03T22:51:51.329608Z","bundle_sha256":"bb95ee425a39a3863889a0918248ac6666a74aa71c119e3fbce13635226cf00e"}}