Dissonance Between Human and Machine Understanding
Open accessThe DOI matches its Crossref record; 27 of the 27 references it cites were found in a registry or an index.
Abstract
Complex machine learning models are deployed in several critical domains including healthcare and autonomous vehicles nowadays, albeit as functional blackboxes. Consequently, there has been a recent surge in interpreting decisions of such complex models in order to explain their actions to humans. Models which correspond to human interpretation of a task are more desirable in certain contexts and can help attribute liability, build trust, expose biases and in turn build better models. It is therefore crucial to understand how and which models conform to human understanding of tasks. In this paper we present a large-scale crowdsourcing study that reveals and quantifies the dissonance between human and machine understanding, through the lens of an image classification task. In particular, we seek to answer the following questions: Which (well performing) complex ML models are closer to humans in their use of features to make accurate predictions? How does task difficulty affect the feature selection capability of machines in comparison to humans? Are humans consistently better at selecting features that make image recognition more accurate? Our findings have important implications on human-machine collaboration, considering that a long term goal in the field of artificial intelligence is to make machines capable of learning and reasoning like humans.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
ImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333
Mastering the game of Go with deep neural networks and tree search
Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., van den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., Lanctot, M., et al. (2016) · Nature · doi:10.1038/nature16961 · Cited by 15,612
“Why Should I Trust You?”: Explaining the Predictions of Any Classifier
Ribeiro, M., Singh, S., & Guestrin, C. (2016) · Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Demonstrations · doi:10.18653/v1/n16-3020 · Cited by 6,682
A perspective on judgment and choice: mapping bounded rationality.
Kahneman, D. (2003) · American Psychologist · doi:10.1037/0003-066x.58.9.697 · Cited by 5,540
Lipton, Z. C. (2018) · Queue · doi:10.1145/3236386.3241340 · Cited by 5,427
Cumulated gain-based evaluation of IR techniques
Järvelin, K., & Kekäläinen, J. (2002) · ACM Transactions on Information Systems · doi:10.1145/582415.582418 · Cited by 4,841 · cited 1×
“egments and the output explanation of SHAP for each neural network model. We analysed the performance of the three neural networks with 3 measures having different semantics: Jaccard Similarity, NDCG [22], weighted Kendall’s �[54] and EMD [52]. The simplest measure is coverage using Jaccard similarity between the huma…”
MethodBuilding machines that learn and think like people
Lake, B. M., Ullman, T. D., Tenenbaum, J. B., & Gershman, S. J. (2017) · Behavioral and Brain Sciences · doi:10.1017/s0140525x16001837 · Cited by 2,550
Building Machines That Learn and Think Like People
Lake, B. M., Ullman, T. D., Tenenbaum, J. B., & Gershman, S. J. (2016) · arXiv · doi:10.48550/arxiv.1604.00289 · Cited by 2,550
Performance-optimized hierarchical models predict neural responses in higher visual cortex
Yamins, D. L. K., Hong, H., Cadieu, C. F., Solomon, E. A., Seibert, D., & DiCarlo, J. J. (2014) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1403112111 · Cited by 1,924
Intelligible Models for HealthCare: Predicting Pneumonia Risk and Hospital 30-day Readmission
Caruana, R., Lou, Y., Gehrke, J., Koch, P., Sturm, M., & Elhadad, N. (2015) · Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining · doi:10.1145/2783258.2788613 · Cited by 1,685
Understanding Black-box Predictions via Influence Functions
Koh, P. W., & Liang, P. (2017) · arXiv · doi:10.48550/arxiv.1703.04730 · Cited by 1,514 · cited 1×
“Model agnostic approaches on the other hand extract post-hoc explanations by treating the original model as a black box either by learning from the output of the black box model, or perturbing the inputs, or both [28, 50].”
BackgroundHuman Decisions and Machine Predictions
Kleinberg, J., Lakkaraju, H., Leskovec, J., Ludwig, J., & Mullainathan, S. (2018) · The Quarterly Journal of Economics · doi:10.1093/qje/qjx032 · Cited by 1,040
Working with Machines: The Impact of Algorithmic and Data-Driven Management on Human Workers
Lee, M. K., Kusbit, D., Metsky, E., & Dabbish, L. (2015) · Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems · doi:10.1145/2702123.2702548 · Cited by 823
Letham, B., Rudin, C., McCormick, T. H., & Madigan, D. (2015) · The Annals of Applied Statistics · doi:10.1214/15-aoas848 · Cited by 766 · cited 1×
“ng a model. Interpretable models can be categorised into two broad classes: model introspective and model agnostic. Model introspection refers to “interpretable” models, such as decision trees, rules [35], additive models [8] and attentionbased networks [70]. Instead of supporting models that are functionally black-bo…”
BackgroundAbdul, A., Vermeulen, J., Wang, D., Lim, B. Y., & Kankanhalli, M. (2018) · Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3173574.3174156 · Cited by 743
Voice Interfaces in Everyday Life
Porcheron, M., Fischer, J. E., Reeves, S., & Sharples, S. (2018) · Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3173574.3174214 · Cited by 717 · cited 1×
“Recent technological advances have led to the growing popularity of a variety of such systems, ranging from voice-based conversational assistants that facilitate and support everyday social interactions [45, 65], mobile health (mHealth) applications which have been proposed to transform healthcare and for health promo…”
BackgroundModel-Agnostic Interpretability of Machine Learning
Ribeiro, M. T., Singh, S., & Guestrin, C. (2016) · arXiv · doi:10.48550/arxiv.1606.05386 · Cited by 699 · cited 1×
“On the other hand there have been approaches that provide post-hoc explanations to already trained models [37, 49].”
Background'It's Reducing a Human Being to a Percentage': Perceptions of Justice in Algorithmic Decisions
Binns, R., Van Kleek, M., Veale, M., Lyngs, U., Zhao, J., & Shadbolt, N. (2018) · Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3173574.3173951 · Cited by 691
Explanations From Intelligent Systems: Theoretical Foundations and Implications for Practice1
Gregor, S., & Benbasat, I. (1999) · MIS Quarterly · doi:10.2307/249487 · Cited by 535 · cited 1×
“Prior works have focused on various attributes of explanations; cognitive fit [18], content type [19], data sources [43], and modality [40].”
BackgroundRight for the Right Reasons: Training Differentiable Models by Constraining their Explanations
Ross, A. S., Hughes, M. C., & Doshi-Velez, F. (2017) · Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2017/371 · Cited by 515 · cited 3×
“Understanding these broad questions are crucial in building machine learning systems [66] and guiding interpretable system design [51].”
BackgroundZarsky, T. (2016) · Science, Technology, & Human Values · doi:10.1177/0162243915605575 · Cited by 422
Explanations as Mechanisms for Supporting Algorithmic Transparency
Rader, E., Cotter, K., & Cho, J. (2018) · Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3173574.3173677 · Cited by 388 · cited 1×
“, algorithmic transparency involves encountering non-obvious information that is typically difficult for the user of a system to learn and experience directly, about how and why a system works the way it does and what this means for the system’s outputs [46].”
BackgroundLocally noisy autonomous agents improve global human coordination in network experiments
Shirado, H., & Christakis, N. A. (2017) · Nature · doi:10.1038/nature22332 · Cited by 288 · cited 1×
“Understanding the full breadth of societal effects that machines can have becomes more complex in hybrid systems composed of many humans and machines interacting; demonstrating collective behaviour [55].”
BackgroundBias in Online Freelance Marketplaces: Evidence from TaskRabbit and Fiverr
Hannák, A., Wagner, C., Garcia, D., Mislove, A., Strohmaier, M., & Wilson, C. (2017) · Proceedings of the 2017 ACM Conference on Computer Supported Cooperative Work and Social Computing · doi:10.1145/2998181.2998327 · Cited by 286
Lee, M. K., & Baykal, S. (2017) · Proceedings of the 2017 ACM Conference on Computer Supported Cooperative Work and Social Computing · doi:10.1145/2998181.2998230 · Cited by 182
Designing and Evaluating mHealth Interventions for Vulnerable Populations: A Systematic Review
Stowell, E., Lyson, M. C., Saksono, H., Wurth, R. C., Jimison, H., Pavel, M., & Parker, A. G. (2018) · Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3173574.3173589 · Cited by 169
Exploring the Role of Conversational Cues in Guided Task Support with Virtual Assistants
Vtyurina, A., & Fourney, A. (2018) · Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3173574.3173782 · Cited by 83
Cited by
A Multidisciplinary Survey and Framework for Design and Evaluation of Explainable AI Systems
Mohseni, S., Zarei, N., & Ragan, E. D. (2021) · ACM Transactions on Interactive Intelligent Systems · doi:10.1145/3387166 · Cited by 785
A Multidisciplinary Survey and Framework for Design and Evaluation of Explainable AI Systems
Mohseni, S., Zarei, N., & Ragan, E. D. (2018) · arXiv · doi:10.48550/arxiv.1811.11839 · Cited by 783
“rch in interpretable machine learning techniques, a core issue in model explanations is in the difference between machine learning model’s decision-making logic and human sense-making as the receiver [90, 199]. ApplicationinUseCase: In our fake news detection case study, the explainable interface design team had previ…”
MethodHuman factors in model interpretability: Industry practices, challenges, and needs
Hong, S. R., Hullman, J., & Bertini, E. (2020) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3392878 · Cited by 219
“…Human-Computer Interaction (HCI) have devoted effort to understanding which ML models can better conform to human understanding of prediction tasks [77], how people’s interaction with ML models affects the quality of decisions [20], and how data scientists use particular tools in conducting their…”
Enhancing AI-Assisted Group Decision Making through LLM-Powered Devil's Advocate
Chiang, C.-W., Lu, Z., Li, Z., & Yin, M. (2024) · Proceedings of the 29th International Conference on Intelligent User Interfaces · doi:10.1145/3640543.3645199 · Cited by 142
Human Reliance on Machine Learning Models When Performance Feedback is Limited: Heuristics and Risks
Lu, Z., & Yin, M. (2021) · Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3411764.3445562 · Cited by 131
Metadata
- Authors
- Zhang, Z., Singh, J., Gadiraju, U., & Anand, A.
- Venue
- Proceedings of the ACM on Human-Computer Interaction
- Published
- 2019-11-07
- Volume
- 3
- Issue
- CSCW
- Pages
- 1-23
- ISSN
- 2573-0142 (ISSN-L)
- OpenAlex
- W2988450373
- Cited by
- 50
- References
- 27
- FWCI
- 3.45
- Type
- journal article
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.