Elamur Research⌕
☰
home / papers / 2024 / ChatGPT Is a Knowledgeable but Inexperienced Solver: An Investigation of Commonsense Problem in Large Language Models

ChatGPT Is a Knowledgeable but Inexperienced Solver: An Investigation of Commonsense Problem in Large Language Models

Paper · 2024 · Ning Bian, Xianpei Han, Le Sun et al.

Summary

Large language models often struggle with commonsense, but it is unclear where the weakness lies: missing knowledge, or failure to apply it. The authors study ChatGPT through four questions. Can it answer commonsense questions? Does it recognise which knowledge a question needs? Does it actually possess that knowledge? Can it use that knowledge when reasoning?

They test on 11 commonsense QA datasets covering 8 domains. The protocol has four stages:

  • direct question answering;
  • asking the model which knowledge is required;
  • having it write out descriptions of that knowledge;
  • answering again with those descriptions as context, compared against "golden" knowledge.

GPT-3 and Instruct GPT serve as baselines. ChatGPT reaches strong accuracy overall but stays weak in some domains and still trails humans. It can produce most of the relevant knowledge when prompted. However, it does not reliably pinpoint which facts a given question requires, and it does not use supplied knowledge well.

The authors therefore describe it as knowledgeable but an inexperienced solver. Future work should focus on better instruction following and commonsense guidance, not just on adding more knowledge.

Key points

  • Benchmarks used: CommonsenseQA, OpenBookQA, WSC, PIQA, Social IQA, ARC (Easy), QASC, HellaSWAG, NumerSense, ProtoQA, MC-TACO. Between them, these datasets test temporal, numerical, scientific, social and physical knowledge, as well as knowledge about events, prototypes and general commonsense.
  • From each development set, 100 questions were sampled; ProtoQA contributed only 52, its full dev set.
  • Models were set up as follows:
    • GPT-3 (davinci) used 4-shot prompting.
    • Instruct GPT (text-davinci-003) and ChatGPT (GPT-3.5 web interface) ran zero-shot with dataset-specific templates.
  • ChatGPT's best results were 94% on ARC and 94.2% on ProtoQA. Instruction-tuned models clearly outperformed GPT-3, which the authors attribute to instruction tuning and human alignment.
  • Although ChatGPT can generate the relevant knowledge, it identifies the specific knowledge each question needs poorly and gains little from knowledge supplied in context.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 9,086 worksArtificial Intelligence · 8,889 worksArts and Humanities · 9,296 worksLanguage and Linguistics · 9,206 works

🧭 Topics 2Research subjects and languages the work is about.Language models & pretraining · 1,343 worksReasoning · 510 works

Cite

Cite as Ning Bian, Xianpei Han, Le Sun, Hongyu Lin, Yaojie Lu, Ben He, Shanshan Jiang and Bin Dong. 2024. ChatGPT Is a Knowledgeable but Inexperienced Solver: An Investigation of Commonsense Problem in Large Language Models. Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024).

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{bian2024chatgpt,
  title = {{ChatGPT Is a Knowledgeable but Inexperienced Solver: An Investigation of Commonsense Problem in Large Language Models}},
  author = {Ning Bian, Xianpei Han, Le Sun, Hongyu Lin, Yaojie Lu, Ben He, Shanshan Jiang and Bin Dong},
  booktitle = {Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)},
  year = {2024},
  doi = {10.63317/32y85i5g9gso},
  url = {https://doi.org/10.63317/32y85i5g9gso}
}

References

59 references, 16 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Shourya Aggarwal, Divyanshu Mandowara, Vishwajeet Agrawal, Dinesh Khandelwal, Parag Singla, and Dinesh Garg. 2021. Explanations for CommonsenseQA: New Dataset and Models. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Confe…
  2. Yejin Bang, Samuel Cahyawijaya, Nayeon Lee, Wenliang Dai, Dan Su, Bryan Wilie, Holy Lovenia, Ziwei Ji, Tiezheng Yu, Willy Chung, et al. 2023. A multitask, multilingual, multimodal evaluation of chatgpt on reasoning, hallucination, and interactivity. ArXiv preprint, abs/2302.04023.
  3. Lisa Bauer, Yicheng Wang, and Mohit Bansal. 2018. Commonsense for generative multi-hop question answering tasks. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 4220–4230, Brussels, Belgium. Association for Computational Linguistics.
  4. David Bender. 2015. Establishing a human baseline for the winograd schema challenge. In MAICS, pages 39–45. find
  5. Prajjwal Bhargava and Vincent Ng. 2022. Commonsense knowledge reasoning and generation with pre-trained language models: A survey. In Thirty-Sixth AAAI Conference on Artificial Intelligence, AAAI 2022, Thirty-Fourth Conference on Innovative Applications of Artificial Intelligence, IAAI 2022, The Twe… find
  6. Yonatan Bisk, Rowan Zellers, Ronan LeBras, Jianfeng Gao, and Yejin Choi. 2020. PIQA: reasoning about physical commonsense in natural language. In The Thirty-Fourth AAAI Conference on Artificial Intelligence, AAAI 2020, The Thirty-Second Innovative Applications of Artificial Intelligence Conference,…
  7. Michael Boratko, Xiang Li, Tim O’Gorman, Rajarshi Das, Dan Le, and Andrew McCallum. 2020. ProtoQA: A question answering dataset for prototypical common-sense reasoning. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 1122–1136, Online. Associa… find
  8. Ronald J. Brachman and Hector J. Levesque. 2022. Toward a new science of common sense. In Thirty-Sixth AAAI Conference on Artificial Intelligence, AAAI 2022, Thirty-Fourth Conference on Innovative Applications of Artificial Intelligence, IAAI 2022, The Twelveth Symposium on Educational Advances in A… find
  9. Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter,…
  10. Tyler A Chang and Benjamin K Bergen. 2023. Language model behavior: A comprehensive survey. ArXiv preprint, abs/2303.11504.
  11. Qianglong Chen, Guohai Xu, Ming Yan, Ji Zhang, Fei Huang, Luo Si, and Yin Zhang. 2023. Distinguish before answer: Generating contrastive explanation as knowledge for commonsense question answering. ArXiv preprint, abs/2305.08135. find
  12. Peter Clark, Isaac Cowhey, Oren Etzioni, Tushar Khot, Ashish Sabharwal, Carissa Schoenick, and Oyvind Tafjord. 2018. Think you have solved question answering? try arc, the ai2 reasoning challenge. ArXiv preprint, abs/1803.05457. find
  13. Anthony G Cohn and Jose Hernandez-Orallo. 2023. Dialectical language model evaluation: An initial appraisal of the commonsense spatial reasoning abilities of llms. ArXiv preprint, abs/2304.11164. find
  14. Leyang Cui, Sijie Cheng, Yu Wu, and Yue Zhang. 2021. On commonsense cues in BERT for solving commonsense tasks. In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 683–693, Online. Association for Computational Linguistics. find
  15. Wanyun Cui and Xingran Chen. 2023. Free lunch for efficient textual commonsense integration in language models. ArXiv preprint, abs/2305.15516. find
  16. Ernest Davis. 2023. Benchmarks for automated commonsense reasoning: A survey. ArXiv preprint, abs/2302.04752. find
  17. Joe Davison, Joshua Feldman, and Alexander Rush. 2019. Commonsense knowledge mining from pretrained models. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 1173–1…
  18. Nick C Ellis. 2008. Implicit and explicit knowledge about language. Encyclopedia of language and education, 6:1–13. find
  19. Jonathan Gordon and Benjamin Van Durme. 2013. Reporting bias and knowledge acquisition. In Proceedings of the 2013 workshop on Automated knowledge base construction, pages 25–30. find
  20. Yuling Gu, Bhavana Dalvi Mishra, and Peter Clark. 2022. Do language models have coherent mental models of everyday things? ArXiv preprint, abs/2212.10029. find
  21. Jie He, Víctor Gutiérrez-Basulto, Jeff Z Pan, et al. 2023. Buca: A binary classification approach to unsupervised commonsense question answering. ArXiv preprint, abs/2305.15932. find
  22. Liwei Jiang, Antoine Bosselut, Chandra Bhagavatula, and Yejin Choi. 2021. “I’m not mad”: Commonsense implications of negation and contradiction. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 43… find
  23. Tushar Khot, Peter Clark, Michal Guerquin, Peter Jansen, and Ashish Sabharwal. 2020. QASC: A dataset for question answering via sentence composition. In The Thirty-Fourth AAAI Conference on Artificial Intelligence, AAAI 2020, The Thirty-Second Innovative Applications of Artificial Intelligence Confe… find
  24. Tassilo Klein and Moin Nabi. 2021. Towards zero-shot commonsense reasoning with self-supervised refinement of language models. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 8737–8743, Online and Punta Cana, Dominican Republic. Association for Comput… find
  25. Kazushi Kondo, Saku Sugawara, and Akiko Aizawa. 2023. Commonsense knowledge transfer for pre-trained language models. ArXiv preprint, abs/2306.02258. find
  26. Md Tahmid Rahman Laskar, M Saiful Bari, Mizanur Rahman, Md Amran Hossen Bhuiyan, Shafiq Joty, and Jimmy Xiangji Huang. 2023. A systematic study and comprehensive evaluation of chatgpt on benchmark datasets. ArXiv preprint, abs/2305.18486.
  27. Hector J Levesque, Ernest Davis, and Leora Morgenstern. 2012. The winograd schema challenge. In Proceedings of the Thirteenth International Conference on Principles of Knowledge Representation and Reasoning, pages 552–561. find
  28. Xiang Lorraine Li, Adhiguna Kuncoro, Cyprien de Masson d’Autume, Phil Blunsom, and Aida Nematzadeh. 2021. Do language models learn commonsense knowledge? abs/2111.00607. — [arXiv:2111.00607](https://arxiv.org/abs/2111.00607) find
  29. ArXiv preprint, Xiang Lorraine Li, Adhiguna Kuncoro, Jordan Hoffmann, Cyprien de Masson d’Autume, Phil Blunsom, and Aida Nematzadeh. 2022. A systematic investigation of commonsense knowledge in large language models. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Proc… find
  30. Bill Yuchen Lin, Seyeon Lee, Rahul Khanna, and Xiang Ren. 2020. Birds have four legs?! NumerSense: Probing Numerical Commonsense Knowledge of Pre-Trained Language Models. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 6862–6868, Online. Assoc…
  31. Hugo Liu and Push Singh. 2004. Conceptnet—a practical commonsense reasoning tool-kit. BT technology journal, 22:211–226. find
  32. Jiacheng Liu, Alisa Liu, Ximing Lu, Sean Welleck, Peter West, Ronan Le Bras, Yejin Choi, and Hannaneh Hajishirzi. 2022. Generated knowledge prompting for commonsense reasoning. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 3… find
  33. Ye Liu, Yao Wan, Lifang He, Hao Peng, and Philip S. Yu. 2021. KG-BART: knowledge graph-augmented BART for generative commonsense reasoning. In Thirty-Fifth AAAI Conference on Artificial Intelligence, AAAI 2021, Thirty-Third Conference on Innovative Applications of Artificial Intelligence, IAAI 2021,… find
  34. Ye Liu, Tao Yang, Zeyu You, Wei Fan, and Philip S. Yu. 2020. Commonsense evidence generation and injection in reading comprehension. In Proceedings of the 21th Annual Meeting of the Special Interest Group on Discourse and Dialogue, pages 61–73, 1st virtual meeting. Association for Computational Ling… find
  35. Kaixin Ma, Filip Ilievski, Jonathan Francis, Satoru Ozaki, Eric Nyberg, and Alessandro Oltramari. 2021. Exploring strategies for generalizable commonsense reasoning with pre-trained models. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 5474–5483, On… find
  36. Todor Mihaylov, Peter Clark, Tushar Khot, and Ashish Sabharwal. 2018. Can a suit of armor conduct electricity? a new dataset for open book question answering. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 2381–2391, Brussels, Belgium. Association fo… find
  37. Yilin Niu, Fei Huang, Jiaming Liang, Wenkai Chen, Xiaoyan Zhu, and Minlie Huang. 2021. A semantic-based method for unsupervised commonsense question answering. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Na… find
  38. Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. Training language models to follow instructions with human feedback. ArXiv preprint, abs/2203.02155.
  39. Ian Porada, Alessandro Sordoni, and Jackie Cheung. 2022. Does pre-training induce systematic inference? how masked language models acquire commonsense knowledge. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Techn… find
  40. Nazneen Fatema Rajani, Bryan McCann, Caiming Xiong, and Richard Socher. 2019. Explain yourself! leveraging language models for commonsense reasoning. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 4932–4942, Florence, Italy. Association for Computat…
  41. Maarten Sap, Hannah Rashkin, Derek Chen, Ronan Le Bras, and Yejin Choi. 2019. Social IQa: Commonsense reasoning about social interactions. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processi… find
  42. Vered Shwartz and Yejin Choi. 2020. Do neural language models overcome reporting bias? In Proceedings of the 28th International Conference on Computational Linguistics, pages 6863–6870, Barcelona, Spain (Online). International Committee on Computational Linguistics. find
  43. Vered Shwartz, Peter West, Ronan Le Bras, Chandra Bhagavatula, and Yejin Choi. 2020. Unsupervised commonsense question answering with self-talk. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 4615–4629, Online. Association for Computational L… find
  44. Yueqing Sun, Yu Zhang, Le Qi, and Qi Shi. 2022. TSGP: Two-stage generative prompting for unsupervised commonsense question answering. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 968–980, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics. find
  45. Alon Talmor, Jonathan Herzig, Nicholas Lourie, and Jonathan Berant. 2019. CommonsenseQA: A question answering challenge targeting commonsense knowledge. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies,…
  46. Alexandre Tamborrino, Nicola Pellicanò, Baptiste Pannier, Pascal Voitot, and Louise Naudin. 2020. Pre-training is (almost) all you need: An application to commonsense reasoning. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 3878–3887, Online. Assoc… find
  47. Xiaojuan Tang, Zilong Zheng, Jiaqi Li, Fanxu Meng, Song-Chun Zhu, Yitao Liang, and Muhan Zhang. 2023. Large language models are in-context semantic reasoners rather than symbolic reasoners. ArXiv preprint, abs/2305.14825. find
  48. Romal Thoppilan, Daniel De Freitas, Jamie Hall, Noam Shazeer, Apoorv Kulshreshtha, Heng-Tze Cheng, Alicia Jin, Taylor Bos, Leslie Baker, Yu Du, et al. 2022. Lamda: Language models for dialog applications. ArXiv preprint, abs/2201.08239.
  49. Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, and Guillaume Lample. 2023. Llama: Open and efficient foundation language models. ArXiv pr…
  50. Peifeng Wang, Filip Ilievski, Muhao Chen, and Xiang Ren. 2021. Do language models perform generalizable commonsense inference? In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 3681–3688, Online. Association for Computational Linguistics. find
  51. Peifeng Wang, Nanyun Peng, Filip Ilievski, Pedro Szekely, and Xiang Ren. 2020. Connecting the dots: A knowledgeable path generator for commonsense question answering. In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 4129–4140, Online. Association for Computational Ling… find
  52. Wangchunshu Zhou, Ronan Le Bras, and Yejin Choi. 2023. Commonsense knowledge transfer for pre-trained language models. ArXiv preprint, abs/2306.02388. find
  53. Xuhui Zhou, Yue Zhang, Leyang Cui, and Dandan Huang. 2020. Evaluating commonsense in pre-trained language models. In The Thirty-Fourth AAAI Conference on Artificial Intelligence, AAAI 2020, The Thirty-Second Innovative Applications of Artificial Intelligence Conference, IAAI 2020, The Tenth AAAI Sym… find
  54. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Ed Chi, Quoc Le, and Denny Zhou. 2022. Chain of thought prompting elicits reasoning in large language models. ArXiv preprint, abs/2201.11903.
  55. Peter West, Chandra Bhagavatula, Jack Hessel, Jena Hwang, Liwei Jiang, Ronan Le Bras, Ximing Lu, Sean Welleck, and Yejin Choi. 2022. Symbolic knowledge distillation: from general language models to commonsense models. In Proceedings of the 2022 Conference of the North American Chapter of the Associa…
  56. Rowan Zellers, Ari Holtzman, Yonatan Bisk, Ali Farhadi, and Yejin Choi. 2019. HellaSwag: Can a machine really finish your sentence? In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 4791–4800, Florence, Italy. Association for Computational Linguistics. find
  57. Yi Zhang, Lei Li, Yunfang Wu, Qi Su, and Xu Sun. 2022. Alleviating the knowledge-language inconsistency: A study for deep commonsense knowledge. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 30:594–604. find
  58. Zirui Zhao, Wee Sun Lee, and David Hsu. 2023. Large language models as commonsense knowledge for large-scale task planning. ArXiv preprint, abs/2305.14078. find
  59. Ben Zhou, Daniel Khashabi, Qiang Ning, and Dan Roth. 2019. “going on a vacation” takes longer than “going for a walk”: A study of temporal commonsense understanding. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference o… find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/bian-2024-chatgpt-is-knowledgeable-but-inexperienced/