skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: Local Justice and Machine Learning: Modeling and Inferring Dynamic Ethical Preferences toward Allocations
We consider a setting in which a social planner has to make a sequence of decisions to allocate scarce resources in a high-stakes domain. Our goal is to understand stakeholders' dynamic moral preferences toward such allocational policies. In particular, we evaluate the sensitivity of moral preferences to the history of allocations and their perceived future impact on various socially salient groups. We propose a mathematical model to capture and infer such dynamic moral preferences. We illustrate our model through small-scale human-subject experiments focused on the allocation of scarce medical resource distributions during a hypothetical viral epidemic. We observe that participants' preferences are indeed history- and impact-dependent. Additionally, our preliminary experimental results reveal intriguing patterns specific to medical resources---a topic that is particularly salient against the backdrop of the global covid-19 pandemic.  more » « less
Award ID(s):
2040929
PAR ID:
10474408
Author(s) / Creator(s):
 ; ; ;
Publisher / Repository:
AAAI
Date Published:
Journal Name:
Proceedings of the AAAI Conference on Artificial Intelligence
Volume:
37
Issue:
5
ISSN:
2159-5399
Page Range / eLocation ID:
5956 to 5964
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; (, Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society)
    Computational preference elicitation methods are tools used to learn people’s preferences quantitatively in a given context. Recent works on preference elicitation advocate for active learning as an efficient method to iteratively construct queries (framed as comparisons between context-specific cases) that are likely to be most informative about an agent’s underlying preferences. In this work, we argue that the use of active learning for moral preference elicitation relies on certain assumptions about the underlying moral preferences, which can be violated in practice. Specifically, we highlight the following common assumptions (a) preferences are stable over time and not sensitive to the sequence of presented queries, (b) the appropriate hypothesis class is chosen to model moral preferences, and (c) noise in the agent’s responses is limited. While these assumptions can be appropriate for preference elicitation in certain domains, prior research on moral psychology suggests they may not be valid for moral judgments. Through a synthetic simulation of preferences that violate the above assumptions, we observe that active learning can have similar or worse performance than a basic random query selection method in certain settings. Yet, simulation results also demonstrate that active learning can still be viable if the degree of instability or noise is relatively small and when the agent’s preferences can be approximately represented with the hypothesis class used for learning. Our study highlights the nuances associated with effective moral preference elicitation in practice and advocates for the cautious use of active learning as a methodology to learn moral preferences. 
    more » « less
  2. ; ; ; (, AEA Papers and Proceedings)
    Many scarce public resources are allocated through wait lists that use priorities for individual agents. A new priority system for allocating deceased donor kidneys was adopted in 2014. This redesign was guided by simulations that held decision-rules fixed. We synthesize recent theoretical results to show that the welfare effects of a mechanism depend on the interaction between dynamic incentives and heterogeneity in preferences. We show evidence suggesting that patient decisions on the deceased donor kidney wait list respond to dynamic incentives. Therefore, an empirical approach to dynamic mechanism design is an essential complement to mechanism design theory in dynamic environments. 
    more » « less
  3. ; (, APHA)
    Background: Across the country, cities increasingly rely on Fire Departments and Emergency Medical Services (EMS) to respond to community members experiencing behavioral health crises, addiction, and homelessness. This increasing scope of work can be hard for first responders, who experience a high rate of burnout and turnover. Methods: Through a four-year civic-partnered participatory action research study, we explored the diverse perspectives of people serving on emergency and street-based response teams in an urban setting in northern California. After formative focus groups, we developed and piloted a slate of interventions to strengthen response programs and prevent moral injury and burnout among first responders. Results: First responders identified a wide range of interventions, ranging from structural systems improvement (e.g. expanded shelter hours, more treatment facilities, access to city-wide data systems) to workplace modifications (e.g. no mandatory overtime, increased casual team-building time, complex call debriefs) that would enhance their ability to serve their communities while preventing burnout and moral injury. We collaboratively and iteratively piloted several interventions and evaluated them using a mixed-methods approach. Through these efforts, we developed a model for how moral injury and burnout arise among first responders using an ecosocial approach. Conclusion: Through this partnered participatory action research study, we found moral injury is a salient concern and that modifiable structural and workplace factors are important drivers of risk. There were also challenges to using participatory methods in the context of a hierarchically structured organization designed for emergency response. We discuss lessons learned through navigating these challenges. 
    more » « less
  4. ; ; ; ; ; ; ; ; ; ; et al (, Proceedings of the Royal Society B: Biological Sciences)
    A wide range of literature connects sex ratio and mating behaviours in non-human animals. However, research examining sex ratio and human mating is limited in scope. Prior work has examined the relationship between sex ratio and desire for short-term, uncommitted mating as well as outcomes such as marriage and divorce rates. Less empirical attention has been directed towards the relationship between sex ratio and mate preferences, despite the importance of mate preferences in the human mating literature. To address this gap, we examined sex ratio's relationship to the variation in preferences for attractiveness, resources, kindness, intelligence and health in a long-term mate across 45 countries ( n = 14 487). We predicted that mate preferences would vary according to relative power of choice on the mating market, with increased power derived from having relatively few competitors and numerous potential mates. We found that each sex tended to report more demanding preferences for attractiveness and resources where the opposite sex was abundant, compared to where the opposite sex was scarce. This pattern dovetails with those found for mating strategies in humans and mate preferences across species, highlighting the importance of sex ratio for understanding variation in human mate preferences. 
    more » « less
  5. (, Judgment and decision making)
    null (Ed.)
    This paper asks whether moral preferences in eight medical dilemmas change as a function of how preferences are expressed, and how people choose when they are faced with two equally attractive help projects. In two large-scale studies, participants first read dilemmas where they “matched” two suggested helping projects (which varied on a single attribute) so that they became equally attractive. They did this by filling in a missing number (e.g., how many male patients must Project M save in order to be equally attractive as Project F which can save 100 female patients). Later, the same participants were asked to choose between the two equally attractive projects. We found robust evidence that people do not choose randomly, but instead tend to choose projects that help female (vs. male), children (vs. adult), innocent (vs. non-innocent), ingroup (vs. outgroup) and existing (vs. future) patients, and imply no (vs. some) risk of a harmful side-effect, even when these projects have been matched as equally attractive as, and save fewer patients than the contrasting project. We also found that some moral preferences are hidden when expressed with matching but apparent when expressed with forced choice. For example, 88–95% of the participants expressed that female and male patients are equally valuable when doing the matching task, but over 80% of them helped female patients in the choice task. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email