Abstract
In many real-world decision-making problems, inputs and data evolve over time, and decisions must be made in this dynamic (online) framework. For example, one might wish to screen resumes in real-time as they are submitted; in this setting, the feedback used to make the screening decisions (such as ultimate hiring decisions on past resumes) might be provided over time as well. While fairness concerns have received significant attention in offline settings, relatively little is known about fairness in these dynamic settings; this begs the question, what does it mean to be fair in online decision-making? This question is complicated because stringent constraints can prevent good decisions later on due to a few bad decisions early on, and further, the data which is used to justify decisions varies over time. Given these challenges, there is a need for new ideas to understand the trade-offs between fairness and utility in dynamic decision-making settings. This chapter surveys existing notions of temporal fairness and reviews recent work in this growing area.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
We chose the term “memory” to allude to the one-sided perspective of the decision-maker in an online setting: the current decision can only reasonably be constrained by previous decisions if the future decisions are unknown. One could also think of any current decision \(x_t\) as imposing a constraint on the future decisions (e.g., a lower bound or an upper bound, especially in the case of fairness across all time). With this interpretation, the term “memory” loses its motivation. While this latter interpretation may be salient from a theoretical perspective of the feasibility of policies, we find the former interpretation to be more compelling in dynamic (online) settings where future decisions are unknown and constraining these is not natural.
- 2.
CS is short for “consistent sequential.”
References
Agarwal A, Foster DP, Hsu D, Kakade SM, Rakhlin A (2013) Stochastic convex optimization with bandit feedback. SIAM J Optim 23(1):213–240
Baker W, Kiewell D, Winkler G (2014) Using big data to make better pricing decisions. McKinsey Analysis
Bansal S, Srivastava A, Arora A (2017) Topic modeling driven content based jobs recommendation engine for recruitment industry. Procedia Comput Sci 122:865–872
Barocas S, Hardt M, Narayanan A (2019) Fairness and machine learning: limitations and opportunities. http://www.fairmlbook.org
Chen N (2021) Multi-armed bandit requiring monotone arm sequences. arXiv:2106.03790
Chouldechova A, Roth A (2018) The frontiers of fairness in machine learning. arXiv:1810.08810
Flaxman AD, Kalai AT, McMahan HB (2005) Online convex optimization in the bandit setting: gradient descent without a gradient. In: Proceedings of the sixteenth annual ACM-SIAM symposium on Discrete algorithms, pp 385–394
Gupta S, Kamble V (2021) Individual fairness in hindsight. J Mach Learn Res 22(144):1–35
Hazan E (2016) Introduction to online convex optimization. Found Trends Optim 2(3–4):157–325
Hazan E, Levy KY (2014) Bandit convex optimization: towards tight bounds. In: NIPS, pp 784–792
Heidari H, Krause A (2018) Preventing disparate treatment in sequential decision making. In: Proceedings of the 27th international joint conference on artificial intelligence, pp 2248–2254
Jagtiani J, Lemieux C (2019) The roles of alternative data and machine learning in fintech lending: evidence from the LendingClub consumer platform. Financ Manag 48(4):1009–1029
Jia S, Li A, Ravi R (2021) Markdown pricing under unknown demand. Available at SSRN 3861379
Kiefer J (1953) Sequential minimax search for a maximum. Proc Am Math Soc 4(3):502–506
Klarsfeld A, Cachat-Rosset G (2021) Equality of treatment, opportunity, and outcomes: mapping the law. In: Oxford research encyclopedia of business and management
Panesar A (2019) Machine learning and AI for healthcare. Springer
Raghavan M, Barocas S, Kleinberg J, Levy K (2020) Mitigating bias in algorithmic hiring: Evaluating claims and practices. In: Proceedings of the 2020 conference on fairness, accountability, and transparency, pp 469–481
Rigano C (2019) Using artificial intelligence to address criminal justice needs. Natl Inst Justice J 280:1–10
Salem J, Gupta S (2023) Secretary problems with biased evaluations using partial ordinal information. Manage Sci
Salem J, Gupta S, Kamble V (2022) Algorithmic challenges in ensuring fairness at the time of decision. arXiv:2103.09287
Shamir O (2013) On the complexity of bandit and derivative-free stochastic convex optimization. In: Conference on learning theory, PMLR, pp 3–24
Spall JC et al (1992) Multivariate stochastic approximation using a simultaneous perturbation gradient approximation. IEEE Trans Autom Control 37(3):332–341
Waters T (2022) Wells fargo bank sued for race discrimination in mortgage lending practices. https://www.usatoday.com/story/money/2022/04/26/wells-fargo-being-sued-discriminating-against-black-borrowers/7451521001/. Published 26 Apr 2022. Accessed 7 Jun 2022
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Institution of Engineers (India)
About this chapter
Cite this chapter
Gupta, S., Kamble, V., Salem, J. (2023). Temporal Fairness in Online Decision-Making. In: Mukherjee, A., Kulshrestha, J., Chakraborty, A., Kumar, S. (eds) Ethics in Artificial Intelligence: Bias, Fairness and Beyond. Studies in Computational Intelligence, vol 1123. Springer, Singapore. https://doi.org/10.1007/978-981-99-7184-8_3
Download citation
DOI: https://doi.org/10.1007/978-981-99-7184-8_3
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-99-7183-1
Online ISBN: 978-981-99-7184-8
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)