{"paper":{"title":"PIQA: Reasoning about Physical Commonsense in Natural Language","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Large pretrained models reach only 77 percent accuracy on physical commonsense questions that humans answer at 95 percent.","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Jianfeng Gao, Ronan Le Bras, Rowan Zellers, Yejin Choi, Yonatan Bisk","submitted_at":"2019-11-26T15:31:46Z","abstract_excerpt":"To apply eyeshadow without a brush, should I use a cotton swab or a toothpick? Questions requiring this kind of physical commonsense pose a challenge to today's natural language understanding systems. While recent pretrained models (such as BERT) have made progress on question answering over more abstract domains - such as news articles and encyclopedia entries, where text is plentiful - in more physical domains, text is inherently limited due to reporting bias. Can AI systems learn to reliably answer physical common-sense questions without experiencing the physical world? In this paper, we in"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"large pretrained models struggle (77%). We provide analysis about the dimensions of knowledge that existing models lack, which offers significant opportunities for future research.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the collected PIQA questions genuinely require physical commonsense reasoning and cannot be solved primarily through linguistic patterns or reporting bias present in the training data.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"PIQA is a new benchmark showing that current AI models achieve 77% on physical commonsense questions versus humans at 95%.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Large pretrained models reach only 77 percent accuracy on physical commonsense questions that humans answer at 95 percent.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"19baf1132d6926b66e49f791e0b60257c465a72c23a1bfb4f3bc723ed74854f3"},"source":{"id":"1911.11641","kind":"arxiv","version":1},"verdict":{"id":"f2f2f679-75c2-4177-9f56-70f1e42efc77","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T14:49:40.257530Z","strongest_claim":"large pretrained models struggle (77%). We provide analysis about the dimensions of knowledge that existing models lack, which offers significant opportunities for future research.","one_line_summary":"PIQA is a new benchmark showing that current AI models achieve 77% on physical commonsense questions versus humans at 95%.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the collected PIQA questions genuinely require physical commonsense reasoning and cannot be solved primarily through linguistic patterns or reporting bias present in the training data.","pith_extraction_headline":"Large pretrained models reach only 77 percent accuracy on physical commonsense questions that humans answer at 95 percent."},"references":{"count":80,"sample":[{"doi":"","year":null,"title":"CVPR , year =","work_id":"374a8d8b-958e-44b8-8c16-270a161b52ae","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"SocialIQA: Commonsense Reasoning about Social Interactions , booktitle =","work_id":"85cba18d-9710-4a1f-baaf-d9afe0f47ee2","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"WINOGRANDE: An Adversarial Winograd Schema Challenge at Scale , author=. AAAI , year=","work_id":"79b330d8-2547-4062-b1ee-339bd722572f","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"ACL , year =","work_id":"054e6f64-f7b0-4ec6-aabe-a7f7c9a335da","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"IROS , year =","work_id":"7604fee5-ff70-49f4-befb-6e4f5799963d","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":80,"snapshot_sha256":"c4efa8c48324514518d8c4444bd98ac02bb57bc92802696c7cc4d12fde74bb29","internal_anchors":4},"formal_canon":{"evidence_count":2,"snapshot_sha256":"185f23d4ef10c48d0939a6322566368657620f70ea6a0f46ecb915e77ccfc765"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}