Book chapter
Aligning to Human Decision-Makers in Military Medical Triage
Case-Based Reasoning Research and Development, v 14775, pp 371-387
24 Jun 2024
Abstract
Expert human decision makers do not make optimal decisions in realistic domains; their decisions are affected by preferences, ethics, background experience, and contextual factors. Often there is no optimal decision, or any consensus on what makes a decision good. In this paper we consider the problem of aligning decisions to human decision makers. To this end, we introduce a novel formulation of an aligned decision-making problem, and present the Trustworthy Algorithmic Delegate (TAD), an integrated AI system that learns to align its decision-making process to target decision-makers using case-based reasoning, Monte Carlo simulation, Bayesian diagnosis, and Naturalistic decision-making. We apply TAD in a military triage domain, where experts make different decisions, and present experimental results showing that it outperforms baselines and ablations at alignment in this domain. Our primary claims are that the combined components of TAD allows for aligned decision-making using a small, learned case base and that TAD outperforms simpler strategies for alignment in this domain.
Metrics
Details
- Title
- Aligning to Human Decision-Makers in Military Medical Triage
- Creators
- Matthew MolineauxRosina O WeberMichael W. FloydDavid MenagerOthalia LarueUrsula AddisonRay KulhanekNoah ReifsnyderChristopher RauchMallika MainaliAnik SenPrateek GoelJustin KarneebJ T TurnerJohn Meyer
- Publication Details
- Case-Based Reasoning Research and Development, v 14775, pp 371-387
- Series
- Lecture Notes in Computer Science
- Publisher
- Springer Nature Switzerland; Cham
- Number of pages
- 17
- Grant note
- Defense Advanced Research Projects Agency (DARPA): FA8650-23-C-7317
We'd like to thank scientists from RTX BBN Technologies and Soar Technology, LLC, who created the profile detection functions and datasets used in our experiments. We'd also like to thank CACI International, Inc., for creating the triage simulation used in our experiments. This research was developed with funding from the Defense Advanced Research Projects Agency (DARPA) In the Moment program, contract number FA8650-23-C-7317. In the Moment (ITM) is assessing whether aligning AI to human values will make humans more willing to delegate decision-making in complex, time-sensitive situations where human experts disagree, such as battlefield triage. The views, opinions and/or findings expressed are those of the authors and should not be interpreted as representing the official views or policies of the Department of Defense or the U.S. Government.
- Resource Type
- Book chapter
- Language
- English
- Academic Unit
- Information Science (Informatics); College of Computing and Informatics
- Web of Science ID
- WOS:001273511900024
- Scopus ID
- 2-s2.0-85198472540
- Other Identifier
- 991021889588904721
InCites Highlights
Data related to this publication, from InCites Benchmarking & Analytics tool:
- Web of Science research areas
- Computer Science, Artificial Intelligence
- Computer Science, Interdisciplinary Applications
- Computer Science, Theory & Methods