Role of Human-AI Interaction in Selective Prediction

被引:0
|
作者
Bondi, Elizabeth [1 ]
Koster, Raphael [2 ]
Sheahan, Hannah [2 ]
Chadwick, Martin [2 ]
Bachrach, Yoram [2 ]
Cemgil, Taylan [2 ]
Paquet, Ulrich [2 ]
Dvijotham, Krishnamurthy [3 ]
机构
[1] Harvard Univ, Cambridge, MA 02138 USA
[2] DeepMind, London, England
[3] Google Brain, Mountain View, CA USA
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Recent work has shown the potential benefit of selective prediction systems that can learn to defer to a human when the predictions of the AI are unreliable, particularly to improve the reliability of AI systems in high-stakes applications like healthcare or conservation. However, most prior work assumes that human behavior remains unchanged when they solve a prediction task as part of a human-AI team as opposed to by themselves. We show that this is not the case by performing experiments to quantify human-AI interaction in the context of selective prediction. In particular, we study the impact of communicating different types of information to humans about the AI system's decision to defer. Using realworld conservation data and a selective prediction system that improves expected accuracy over that of the human or AI system working individually, we show that this messaging has a significant impact on the accuracy of human judgements. Our results study two components of the messaging strategy: 1) Whether humans are informed about the prediction of the AI system and 2) Whether they are informed about the decision of the selective prediction system to defer. By manipulating these messaging components, we show that it is possible to significantly boost human performance by informing the human of the decision to defer, but not revealing the prediction of the AI. We therefore show that it is vital to consider how the decision to defer is communicated to a human when designing selective prediction systems, and that the composite accuracy of a human-AI team must be carefully evaluated using a human-in-the-loop framework.
引用
收藏
页码:5286 / 5294
页数:9
相关论文
共 50 条
  • [1] Human-AI interaction
    Sun, Yongqiang
    Shen, Xiao-Liang
    Zhang, Kem Z.K.
    [J]. Data and Information Management, 2023, 7 (03)
  • [2] Human-AI Interaction and AI Avatars
    Liu, Yuxin
    Siau, Keng L.
    [J]. HCI INTERNATIONAL 2023 LATE BREAKING PAPERS, HCII 2023, PT VI, 2023, 14059 : 120 - 130
  • [3] Guidelines for Human-AI Interaction
    Amershi, Saleema
    Weld, Dan
    Vorvoreanu, Mihaela
    Fourney, Adam
    Nushi, Besmira
    Collisson, Penny
    Suh, Jina
    Iqbal, Shamsi
    Bennett, Paul N.
    Inkpen, Kori
    Teevan, Jaime
    Kikin-Gil, Ruth
    Horvitz, Eric
    [J]. CHI 2019: PROCEEDINGS OF THE 2019 CHI CONFERENCE ON HUMAN FACTORS IN COMPUTING SYSTEMS, 2019,
  • [4] Assessing Human-AI Interaction Early through Factorial Surveys: A Study on the Guidelines for Human-AI Interaction
    Li, Tianyi
    Vorvoreanu, Mihaela
    DeBellis, Derek
    Amershi, Saleema
    [J]. ACM TRANSACTIONS ON COMPUTER-HUMAN INTERACTION, 2023, 30 (05)
  • [5] The role of perceived freewill in crises of human-AI interaction: the mediating role of ethical responsibility of AI
    Ahn, Jungyong
    Kim, Jungwon
    Sung, Yongjun
    [J]. INTERNATIONAL JOURNAL OF ADVERTISING, 2024, 43 (05) : 847 - 873
  • [6] Theory of Mind in Human-AI Interaction
    Wang, Qiaosi
    Walsh, Sarah E.
    Si, Mei
    Kephart, Jeffrey O.
    Weisz, Justin D.
    Goel, Ashok K.
    [J]. EXTENDED ABSTRACTS OF THE 2024 CHI CONFERENCE ON HUMAN FACTORS IN COMPUTING SYSTEMS, CHI 2024, 2024,
  • [7] Interpretability as a dynamic of human-AI interaction
    Thieme, Anja
    Cutrell, Ed
    Morrison, Cecily
    Taylor, Alex
    Sellen, Abigail
    [J]. Interactions, 2020, 27 (05): : 40 - 45
  • [8] Human-aware AI -A foundational framework for human-AI interaction
    Sreedharan, Sarath
    [J]. AI MAGAZINE, 2023, 44 (04) : 460 - 466
  • [9] Human-AI Interaction: Human Behavior Routineness Shapes AI Performance
    Shandong University, School of Software, Jinan
    250100, China
    不详
    CA
    94086, United States
    [J]. IEEE Trans Knowl Data Eng, 2024, 12 (8476-8487): : 8476 - 8487
  • [10] Human-AI Interaction in Symbolic Problem Solving
    Jones, Benjamin T.
    [J]. 2018 IEEE SYMPOSIUM ON VISUAL LANGUAGES AND HUMAN-CENTRIC COMPUTING (VL/HCC), 2018, : 265 - 266