IDEAS home Printed from https://ideas.repec.org/p/arx/papers/2012.02394.html
   My bibliography  Save this paper

Biased Programmers? Or Biased Data? A Field Experiment in Operationalizing AI Ethics

Author

Listed:
  • Bo Cowgill
  • Fabrizio Dell'Acqua
  • Samuel Deng
  • Daniel Hsu
  • Nakul Verma
  • Augustin Chaintreau

Abstract

Why do biased predictions arise? What interventions can prevent them? We evaluate 8.2 million algorithmic predictions of math performance from $\approx$400 AI engineers, each of whom developed an algorithm under a randomly assigned experimental condition. Our treatment arms modified programmers' incentives, training data, awareness, and/or technical knowledge of AI ethics. We then assess out-of-sample predictions from their algorithms using randomized audit manipulations of algorithm inputs and ground-truth math performance for 20K subjects. We find that biased predictions are mostly caused by biased training data. However, one-third of the benefit of better training data comes through a novel economic mechanism: Engineers exert greater effort and are more responsive to incentives when given better training data. We also assess how performance varies with programmers' demographic characteristics, and their performance on a psychological test of implicit bias (IAT) concerning gender and careers. We find no evidence that female, minority and low-IAT engineers exhibit lower bias or discrimination in their code. However, we do find that prediction errors are correlated within demographic groups, which creates performance improvements through cross-demographic averaging. Finally, we quantify the benefits and tradeoffs of practical managerial or policy interventions such as technical advice, simple reminders, and improved incentives for decreasing algorithmic bias.

Suggested Citation

  • Bo Cowgill & Fabrizio Dell'Acqua & Samuel Deng & Daniel Hsu & Nakul Verma & Augustin Chaintreau, 2020. "Biased Programmers? Or Biased Data? A Field Experiment in Operationalizing AI Ethics," Papers 2012.02394, arXiv.org.
  • Handle: RePEc:arx:papers:2012.02394
    as

    Download full text from publisher

    File URL: http://arxiv.org/pdf/2012.02394
    File Function: Latest version
    Download Restriction: no
    ---><---

    References listed on IDEAS

    as
    1. J. Aislinn Bohren & Kareem Haggag & Alex Imas & Devin G. Pope, 2019. "Inaccurate Statistical Discrimination: An Identification Problem," PIER Working Paper Archive 19-010, Penn Institute for Economic Research, Department of Economics, University of Pennsylvania, revised 17 Jul 2020.
    2. Phelps, Edmund S, 1972. "The Statistical Theory of Racism and Sexism," American Economic Review, American Economic Association, vol. 62(4), pages 659-661, September.
    Full references (including those not matched with items on IDEAS)

    Citations

    Citations are extracted by the CitEc Project, subscribe to its RSS feed for this item.
    as


    Cited by:

    1. Xiyang Hu & Yan Huang & Beibei Li & Tian Lu, 2022. "Uncovering the Source of Machine Bias," Papers 2201.03092, arXiv.org.
    2. Bessen, James & Impink, Stephen Michael & Reichensperger, Lydia & Seamans, Robert, 2022. "The role of data for AI startup growth," Research Policy, Elsevier, vol. 51(5).
    3. Mallory Avery & Andreas Leibbrandt & Joseph Vecci, 2023. "Does Artificial Intelligence Help or Hurt Gender Diversity? Evidence from Two Field Experiments on Recruitment in Tech," Monash Economics Working Papers 2023-09, Monash University, Department of Economics.
    4. Tanvir Ahmed Khan, 2023. "Can Unbiased Predictive AI Amplify Bias?," Working Paper 1510, Economics Department, Queen's University.
    5. Xiang Hui & Oren Reshef & Luofeng Zhou, 2023. "The Short-Term Effects of Generative Artificial Intelligence on Employment: Evidence from an Online Labor Market," CESifo Working Paper Series 10601, CESifo.

    Most related items

    These are the items that most often cite the same works as this one and are cited by the same works as this one.
    1. J. Michelle Brock & Ralph De Haas, 2023. "Discriminatory Lending: Evidence from Bankers in the Lab," American Economic Journal: Applied Economics, American Economic Association, vol. 15(2), pages 31-68, April.
    2. Daniel Martin & Philip Marx, 2022. "A Robust Test of Prejudice for Discrimination Experiments," Management Science, INFORMS, vol. 68(6), pages 4527-4536, June.
    3. Barron, Kai & Ditlmann, Ruth & Gehrig, Stefan & Schweighofer-Kodritsch, Sebastian, 2020. "Explicit and implicit belief-based gender discrimination: A hiring experiment," Discussion Papers, Research Unit: Economics of Change SP II 2020-306, WZB Berlin Social Science Center.
    4. Achard, Pascal & Suetens, Sigrid, 2023. "The Causal Effect of Ethnic Diversity on Support for Redistribution and the Role of Discrimination," Other publications TiSEM a5e6e0cd-5e07-4a24-a15c-a, Tilburg University, School of Economics and Management.
    5. Cordoba, Juan C. & Isojärvi, Anni & Li, Haoran, 2020. "Equilibrium Unemployment: The Role Of Discrimination," ISU General Staff Papers 202011140800001116, Iowa State University, Department of Economics.
    6. Suchon, Rémi & Houser, Daniel, 2022. "Image spillovers in groups and misreporting," Journal of Economic Behavior & Organization, Elsevier, vol. 198(C), pages 302-314.
    7. D'Acunto, Francesco & Ghosh, Pulak & Jain, Rajiv & Rossi, Alberto G., 2022. "How costly are cultural biases?," LawFin Working Paper Series 34, Goethe University, Center for Advanced Studies on the Foundations of Law and Finance (LawFin).
    8. Ruzzier, Christian A. & Woo, Marcelo D., 2023. "Discrimination with inaccurate beliefs and confirmation bias," Journal of Economic Behavior & Organization, Elsevier, vol. 210(C), pages 379-390.
    9. Forth, John & Theodoropoulos, Nikolaos, 2022. "Earnings Discrimination in the Workplace," IZA Discussion Papers 15357, Institute of Labor Economics (IZA).
    10. Halla, Martin & Kah, Christopher & Sausgruber, Rupert, 2021. "Testing for Ethnic Discrimination in Outpatient Health Care: Evidence from a Field Experiment in Germany," Department of Economics Working Paper Series 319, WU Vienna University of Economics and Business.
    11. Asad, Sher Afghan & Banerjee, Ritwik & Bhattacharya, Joydeep, 2020. "Do workers discriminate against their out-group employers? Evidence from the gig economy," ISU General Staff Papers 202002230800001098, Iowa State University, Department of Economics.
    12. David Arnold & Will Dobbie & Peter Hull, 2022. "Measuring Racial Discrimination in Bail Decisions," American Economic Review, American Economic Association, vol. 112(9), pages 2992-3038, September.
    13. Felipe Goncalves & Steven Mello, 2021. "A Few Bad Apples? Racial Bias in Policing," American Economic Review, American Economic Association, vol. 111(5), pages 1406-1441, May.
    14. Adair Morse & Karen Pence, 2021. "Technological Innovation and Discrimination in Household Finance," Springer Books, in: Raghavendra Rau & Robert Wardrop & Luigi Zingales (ed.), The Palgrave Handbook of Technological Finance, pages 783-808, Springer.
    15. Broockman, David E. & Soltas, Evan J., 2020. "A natural experiment on discrimination in elections," Journal of Public Economics, Elsevier, vol. 188(C).
    16. Miguel A. Fonseca & Ashley McCrea, 2023. "The role of shortlisting in shifting gender beliefs on performance: experimental evidence," Discussion Papers 2315, University of Exeter, Department of Economics.
    17. Ayaita, Adam, 2021. "Labor Market Discrimination and Statistical Differences in Unobserved Characteristics of Applicants," EconStor Preprints 236615, ZBW - Leibniz Information Centre for Economics.
    18. Oliver Gürtler & Lennart Struth, 2021. "Do Workers Benefit from Wage Transparency Rules?," ECONtribute Discussion Papers Series 105, University of Bonn and University of Cologne, Germany.
    19. Sona Badalyan & Darya Korlyakova & Rastislav Rehak, 2023. "Disclosure Discrimination: An Experiment Focusing on Communication in the Hiring Process," CERGE-EI Working Papers wp743, The Center for Economic Research and Graduate Education - Economics Institute, Prague.
    20. Arthur Charpentier, 2022. "Quantifying fairness and discrimination in predictive models," Papers 2212.09868, arXiv.org.

    More about this item

    NEP fields

    This paper has been announced in the following NEP Reports:

    Statistics

    Access and download statistics

    Corrections

    All material on this site has been provided by the respective publishers and authors. You can help correct errors and omissions. When requesting a correction, please mention this item's handle: RePEc:arx:papers:2012.02394. See general information about how to correct material in RePEc.

    If you have authored this item and are not yet registered with RePEc, we encourage you to do it here. This allows to link your profile to this item. It also allows you to accept potential citations to this item that we are uncertain about.

    If CitEc recognized a bibliographic reference but did not link an item in RePEc to it, you can help with this form .

    If you know of missing items citing this one, you can help us creating those links by adding the relevant references in the same way as above, for each refering item. If you are a registered author of this item, you may also want to check the "citations" tab in your RePEc Author Service profile, as there may be some citations waiting for confirmation.

    For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: arXiv administrators (email available below). General contact details of provider: http://arxiv.org/ .

    Please note that corrections may take a couple of weeks to filter through the various RePEc services.

    IDEAS is a RePEc service. RePEc uses bibliographic data supplied by the respective publishers.