CoChat

IS AI GROUND TRUTH REALLY TRUE? THE DANGERS OF TRAINING AND EVALUATING AI TOOLS BASED ON EXPERTS’ KNOW-WHAT.

Lebovitz, Sarah, Levina, Natalia, Lifshitz-Assaf, Hila · 2021 · MIS Quarterly

The DOI matches its Crossref record; 64 of the 64 references it cites were found in a registry or an index.

Abstract

Organizational decision-makers need to evaluate AI tools in light of increasing claims that such tools outperform human experts. Yet, measuring the quality of knowledge work is challenging, raising the question of how to evaluate AI performance in such contexts. We investigate this question through a field study of a major U.S. hospital, observing how managers evaluated five different machine-learning (ML) based AI tools. Each tool reported high performance according to standard AI accuracy measures, which were based on ground truth labels provided by qualified experts. Trying these tools out in practice, however, revealed that none of them met expectations. Searching for explanations, managers began confronting the high uncertainty of experts’ know-what knowledge captured in ground truth labels used to train and validate ML models. In practice, experts address this uncertainty by drawing on rich know-how practices, which were not incorporated into these ML-based tools. Discovering the disconnect between AI’s know-what and experts’ know-how enabled managers to better understand the risks and benefits of each tool. This study shows dangers of treating ground truth labels used in ML models objectively when the underlying knowledge is uncertain. We outline implications of our study for developing, training, and evaluating AI for knowledge work.

Overview

Integrity signals

Registry

Crossref · 10.25300/misq/2021/16564

Cites

64 of 64 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Verification

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Citations

Cites

Research works printed but not linked: 33

  1. Bechky, 2021, Blood, Powder, and Residue: How Crime Labs Translate Evidence into Proof
  2. Bechmann, 2019, Unsupervised by Any Other Name: Hidden Layers of Knowledge Production in Artificial Intelligence on Social Media, 6
  3. Berger, 1966, The Social Construction of Reality: A Treatise in the Sociology of Knowledge
  4. Bowker, 2000, Sorting Things Out: Classification and Its Consequences
  5. Chandrasekaran, 1980, RADEX– Towards a Computer-Based Radiology Consultant, 463
  6. Charmaz, 2014, Constructing Grounded Theory
  7. Dhar, 2016, When to Trust Robots with Decisions, and When Not To, Harvard Business Review
  8. Dreyfus, 2000, Mind Over Machine
  9. Garud, 1997, On the Distinction between Know-How, KnowWhat, and Know-Why, Advances in Strategic Management, 14, 81
  10. Glaser, 1967, Discovering Grounded Theory
  11. Langlotz, 2019, Will Artificial Intelligence Replace Radiologists?, Radiology: Artificial Intelligence, 1, e190058
  12. Latour, 1987, Science in Action: How to Follow Scientists and Engineers through Society
  13. Leonard-Barton, 1995, Wellsprings of Knowledge: Building and Sustaining the Sources of Innovation
  14. Mongan, 2020, Checklist for Artificial Intelligence in Medical Imaging (CLAIM): A Guide for Authors and Reviewers, Radiology: Artificial Intelligence, 2, e200029
  15. Nicolini, 2012, Practice Theory, Work, and Organization: An Introduction
  16. Oakden-Rayner, 2019, The Rebirth of CAD: How Is Modern AI Different from the CAD We Know?, Radiology: Artificial Intelligence, 1
  17. Pinch, 1987, The Social Construction of Facts and Artifacts: Or How the Sociology of Science and the Sociology of Technology Might Benefit Each Other, 17
  18. Polanyi, 1958, Personal Knowledge: Towards a Post-Critical Philosophy
  19. Polanyi, 1966, The Tacit Dimension
  20. Provost, 2013, Data Science for Business: What You Need to Know about Data Mining and Data-Analytic Thinking
  21. Rao, 2017, Sizing the Prize: What’s the Real Value of AI for Your Business and How Can You Capitalise
  22. Ryle, 1949, The Concept of Mind
  23. Schön, 1983, The Reflective Practitioner: How Professionals Think in Action
  24. Seamans, 2019, AI and the Economy, Innovation Policy and the Economy, 19, 161
  25. Shoham, 2018, The AI Index 2018 Annual Report
  26. Simon, 1987, Making Management Decisions: The Role of Intuition and Emotion, The Academy of Management Executive (1987-1989), 1, 57
  27. Smyth, 1994, Inferring Ground Truth from Subjective Labelling of Venus Images, Advances in Neural Information Processing Systems, 1085
  28. Spradley, 1979, The Ethnographic Interview
  29. Star, 1995, Ecologies of Knowledge: Work and Politics in Science and Technology
  30. Suchman, 1987, Plans and Situated Actions
  31. Timmermans, 2003, The Gold Standard: The Challenge of Evidence-Based Medicine
  32. Van Maanen, 1998, Qualitative Studies of Organizations
  33. von Hippel, 1988, The Sources of Innovation

Cited by

See all 588

Metadata

Authors
Lebovitz, S., Levina, N., & Lifshitz-Assaf, H.
Affiliations
  • University of Virginia (US)
  • New York University (US)
  • New York University, Stern School of Business, New York, NY 10003 U.S.A.
Venue
MIS Quarterly
Publisher
MIS Quarterly
Published
2021-09-01
Volume
45
Issue
3
Pages
1501-1526
ISSN
  • 0276-7783 (ISSN-L)
  • 2162-9730
OpenAlex
W3168956186
Cited by
49
References
59
Citations by year
2026: 3 · 2025: 11 · 2024: 16 · 2023: 14 · 2022: 3 · 2021: 1
FWCI
3.01
Type
journal article
Language
en
Open access
closed
Locations
  • MIS Quarterly (publishedVersion)
  • Warwick Research Archive Portal (University of Warwick) (acceptedVersion)
  • Journal of the Association for Information Systems (submittedVersion)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence Applications
  • Social Sciences › Business, Management and Accounting › Management Information Systems › Information Systems Theories and Implementation
  • Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
Keywords
AI evaluation, ground truth labels, expert knowledge, know-how, knowledge work, machine learning, hospital management, uncertainty, risks of AI adoption, human-AI comparison, organizational decision-making

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.