Polite AI mitigates user susceptibility to AI hallucinations.

IF 2 3区 工程技术 Q3 ENGINEERING, INDUSTRIAL
Richard Pak, Ericka Rovira, Anne McLaughlin
{"title":"Polite AI mitigates user susceptibility to AI hallucinations.","authors":"Richard Pak, Ericka Rovira, Anne McLaughlin","doi":"10.1080/00140139.2024.2434604","DOIUrl":null,"url":null,"abstract":"<p><p>With their increased capability, AI-based chatbots have become increasingly popular tools to help users answer complex queries. However, these chatbots may hallucinate, or generate incorrect but very plausible-sounding information, more frequently than previously thought. Thus, it is crucial to examine strategies to mitigate human susceptibility to hallucinated output. In a between-subjects experiment, participants completed a difficult quiz with assistance from either a polite or neutral-toned AI chatbot, which occasionally provided hallucinated (incorrect) information. Signal detection analysis revealed that participants interacting with polite-AI showed modestly higher sensitivity in detecting hallucinations and a more conservative response bias compared to those interacting with neutral-toned AI. While the observed effect sizes were modest, even small improvements in users' ability to detect AI hallucinations can have significant consequences, particularly in high-stakes domains or when aggregated across millions of AI interactions.</p>","PeriodicalId":50503,"journal":{"name":"Ergonomics","volume":" ","pages":"1-11"},"PeriodicalIF":2.0000,"publicationDate":"2024-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Ergonomics","FirstCategoryId":"5","ListUrlMain":"https://doi.org/10.1080/00140139.2024.2434604","RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"ENGINEERING, INDUSTRIAL","Score":null,"Total":0}
引用次数: 0

Abstract

With their increased capability, AI-based chatbots have become increasingly popular tools to help users answer complex queries. However, these chatbots may hallucinate, or generate incorrect but very plausible-sounding information, more frequently than previously thought. Thus, it is crucial to examine strategies to mitigate human susceptibility to hallucinated output. In a between-subjects experiment, participants completed a difficult quiz with assistance from either a polite or neutral-toned AI chatbot, which occasionally provided hallucinated (incorrect) information. Signal detection analysis revealed that participants interacting with polite-AI showed modestly higher sensitivity in detecting hallucinations and a more conservative response bias compared to those interacting with neutral-toned AI. While the observed effect sizes were modest, even small improvements in users' ability to detect AI hallucinations can have significant consequences, particularly in high-stakes domains or when aggregated across millions of AI interactions.

礼貌的AI减轻了用户对AI幻觉的敏感性。
随着其功能的增强,基于人工智能的聊天机器人已经成为帮助用户回答复杂问题的越来越受欢迎的工具。然而,这些聊天机器人可能会产生幻觉,或者产生不正确但听起来非常可信的信息,比之前想象的要频繁。因此,研究减轻人类对幻觉输出的敏感性的策略是至关重要的。在一项受试者之间的实验中,参与者在礼貌或中性的人工智能聊天机器人的帮助下完成了一项困难的测试,人工智能聊天机器人偶尔会提供幻觉(不正确的)信息。信号检测分析显示,与与中性调的人工智能互动的参与者相比,与礼貌型人工智能互动的参与者在检测幻觉方面表现出更高的敏感度,并且反应偏差更保守。虽然观察到的效应大小不大,但即使用户检测人工智能幻觉的能力有微小的提高,也会产生重大的后果,特别是在高风险领域,或者在数百万人工智能交互的汇总中。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
Ergonomics
Ergonomics 工程技术-工程:工业
CiteScore
4.60
自引率
12.50%
发文量
147
审稿时长
6 months
期刊介绍: Ergonomics, also known as human factors, is the scientific discipline that seeks to understand and improve human interactions with products, equipment, environments and systems. Drawing upon human biology, psychology, engineering and design, Ergonomics aims to develop and apply knowledge and techniques to optimise system performance, whilst protecting the health, safety and well-being of individuals involved. The attention of ergonomics extends across work, leisure and other aspects of our daily lives. The journal Ergonomics is an international refereed publication, with a 60 year tradition of disseminating high quality research. Original submissions, both theoretical and applied, are invited from across the subject, including physical, cognitive, organisational and environmental ergonomics. Papers reporting the findings of research from cognate disciplines are also welcome, where these contribute to understanding equipment, tasks, jobs, systems and environments and the corresponding needs, abilities and limitations of people. All published research articles in this journal have undergone rigorous peer review, based on initial editor screening and anonymous refereeing by independent expert referees.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信