Bansal et al. (2021)
paperAuthors
Credibility Rating
Good quality. Reputable source with community review or editorial standards, but less rigorous than peer-reviewed venues.
Rating inherited from publication venue: arXiv
Empirical study examining human overreliance on AI-powered decision support systems and the ineffectiveness of explanations in mitigating this behavior, directly relevant to AI safety concerns about human-AI interaction and appropriate reliance calibration.
Paper Details
Metadata
Abstract
People supported by AI-powered decision support tools frequently overrely on the AI: they accept an AI's suggestion even when that suggestion is wrong. Adding explanations to the AI decisions does not appear to reduce the overreliance and some studies suggest that it might even increase it. Informed by the dual-process theory of cognition, we posit that people rarely engage analytically with each individual AI recommendation and explanation, and instead develop general heuristics about whether and when to follow the AI suggestions. Building on prior research on medical decision-making, we designed three cognitive forcing interventions to compel people to engage more thoughtfully with the AI-generated explanations. We conducted an experiment (N=199), in which we compared our three cognitive forcing designs to two simple explainable AI approaches and to a no-AI baseline. The results demonstrate that cognitive forcing significantly reduced overreliance compared to the simple explainable AI approaches. However, there was a trade-off: people assigned the least favorable subjective ratings to the designs that reduced the overreliance the most. To audit our work for intervention-generated inequalities, we investigated whether our interventions benefited equally people with different levels of Need for Cognition (i.e., motivation to engage in effortful mental activities). Our results show that, on average, cognitive forcing interventions benefited participants higher in Need for Cognition more. Our research suggests that human cognitive motivation moderates the effectiveness of explainable AI solutions.
Summary
This paper addresses the problem of overreliance on AI decision support systems, where users accept AI suggestions even when incorrect. The authors find that simple explanations do not reduce overreliance and may increase it. They propose three cognitive forcing interventions designed to compel users to engage more thoughtfully with AI explanations, drawing on dual-process theory and medical decision-making research. In an experiment with 199 participants, cognitive forcing significantly reduced overreliance compared to simple explainable AI approaches, though users rated these interventions less favorably. Importantly, the interventions benefited participants with higher Need for Cognition more, suggesting that individual differences in cognitive motivation moderate the effectiveness of explainable AI solutions.
Cited by 1 page
| Page | Type | Quality |
|---|---|---|
| AI-Human Hybrid Systems | Approach | 91.0 |
Cached Content Preview
[2102.09692] To Trust or to Think: Cognitive Forcing Functions Can Reduce Overreliance on AI in AI-assisted Decision-making
To Trust or to Think: Cognitive Forcing Functions Can Reduce Overreliance on AI in AI-assisted Decision-making
Zana Buçinca
zbucinca@seas.harvard.edu
Harvard University 33 Oxford St. Cambridge MA 02138 USA
,
Maja Barbara Malaya
217723@edu.p.lodz.pl
Lodz University of Technology ul. Stefana Żeromskiego 116 90-924 Łódź Poland
and
Krzysztof Z. Gajos
kgajos@eecs.harvard.edu
Harvard University 33 Oxford St. Cambridge MA 02138 USA
(October 2020; January 2021; January 2021)
Abstract.
People supported by AI-powered decision support tools frequently overrely on the AI: they accept an AI’s suggestion even when that suggestion is wrong. Adding explanations to the AI decisions does not appear to reduce the overreliance and some studies suggest that it might even increase it.
Informed by the dual-process theory of cognition, we posit that people rarely engage analytically with each individual AI recommendation and explanation, and instead develop general heuristics about whether and when to follow the AI suggestions. Building on prior research on medical decision-making, we designed three cognitive forcing interventions to compel people to engage more thoughtfully with the AI-generated explanations.
We conducted an experiment (N=199), in which we compared our three cognitive forcing designs to two simple explainable AI approaches and to a no-AI baseline. The results demonstrate that cognitive forcing significantly reduced overreliance compared to the simple explainable AI approaches. However, there was a trade-off: people assigned the least favorable subjective ratings to the designs that reduced the overreliance the most. To audit our work for intervention-generated inequalities, we investigated whether our interventions benefited equally people with different levels of Need for Cognition (i.e., motivation to engage in effortful mental activities). Our results show that, on average, cognitive forcing interventions benefited participants higher in Need for Cognition more.
Our research suggests that human cognitive motivation moderates the effectiveness of explainable AI solutions.
explanations; artificial intelligence; trust; cognition
† † booktitle: Woodstock ’18: ACM Symposium on Neural Gaze Detection,
June 03–05, 2018, Woodstock, NY † † copyright: acmlicensed † † journal: PACMHCI † † journalyear: 2021 † † journalvolume: 5 † † journalnumber: CSCW1 † † article: 188 † † publicationmonth: 4 † † price: 15.00 † † doi: 10.1145/3449287 † † ccs: Human-centered computing Interaction design
1. Introduction
From loan approval to disease diagnosis, humans are increasingly being assisted by artificially intelligent (AI) systems in decision-making tasks. By combining two types of intelligence, these emerging sociotechnical systems (i.e
... (truncated, 89 KB total)fa89fdbc996108aa | Stable ID: sid_S28201wv49