Drivers are blamed more than their automated cars when both make mistakes

Awad, Edmond; Levine, Sydney; Kleiman-Weiner, Max; Dsouza, Sohan; Tenenbaum, Joshua B.; Shariff, Azim; Bonnefon, Jean‐François; Rahwan, Iyad

doi:10.1038/s41562-019-0762-8

Cited by 95 publications

(101 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the last 5 years, the development of artificial intelligence (AI) has leapfrogged forward across several industries, ranging from quality control in various manufacturing areas to improved automation of production processes and enhanced diagnostics in medical applications. 1 Examples include voice and facial recognition, 2,3 landmark localization, 4 autonomic driving, [5][6][7][8][9] and a wide array of medical imaging modalities. [10][11][12][13][14][15][16][17][18][19][20][21] From a clinical standpoint, demonstrating the application of AI and its deep neural network learning algorithms is highly relevant and timely due to the ongoing debate on the necessity of advanced medical and surgical intervention while costs are rising.…”

Section: Introductionmentioning

confidence: 99%

Feasibility of Deep Learning Algorithms for Reporting in Routine Spine Magnetic Resonance Imaging

Lewandrowski

Muraleedharan²,

Eddy³

et al. 2020

Int J Spine Surg

View full text Add to dashboard Cite

Background: Artificial intelligence is gaining traction in automated medical imaging analysis. Development of more accurate magnetic resonance imaging (MRI) predictors of successful clinical outcomes is necessary to better define indications for surgery, improve clinical outcomes with targeted minimally invasive and endoscopic procedures, and realize cost savings by avoiding more invasive spine care.Objective: To demonstrate the ability for deep learning neural network models to identify features in MRI DICOM datasets that represent varying intensities or severities of common spinal pathologies and injuries and to demonstrate the feasibility of generating automated verbal MRI reports comparable to those produced by reading radiologists.Methods: A 3-dimensional (3D) anatomical model of the lumbar spine was fitted to each of the patient's MRIs by a team of technicians. MRI T1, T2, sagittal, axial, and transverse reconstruction image series were used to train segmentation models by the intersection of the 3D model through these image sequences. Class definitions were extracted from the radiologist report for the central canal: (0) no disc bulge/protrusion/canal stenosis, (1) disc bulge without canal stenosis, (2) disc bulge resulting in canal stenosis, and (3) disc herniation/protrusion/extrusion resulting in canal stenosis. Both the left and right neural foramina were assessed with either (0) neural foraminal stenosis absent, or (1) neural foramina stenosis present. Reporting criteria for the pathologies at each disc level and, when available, the grading of severity were extracted, and a natural language processing model was used to generate a verbal and written report. These data were then used to train a set of very deep convolutional neural network models, optimizing for minimal binary cross-entropy for each classification.Results: The initial prediction validation of the implemented deep learning algorithm was done on 20% of the dataset, which was not used for artificial intelligence training. Of the 17,800 total disc locations for which MRI images and radiology reports were available, 14,720 were used to train the model, and 3560 were used to validate against. The convergence of validation accuracy achieved with the deep learning algorithm for the foraminal stenosis detector was 81% (sensitivity ¼ 72.4.4%, specificity ¼ 83.1%) after 25 complete iterations through the entire training dataset (epoch).The accuracy was 86.2% (sensitivity ¼ 91.1%, specificity ¼ 82.5%) for the central stenosis detector and 85.2% (sensitivity ¼ 81.8%, specificity ¼ 87.4%) for the disc herniation detector.Conclusions: Deep learning algorithms may be used for routine reporting in spine MRI. There was a minimal disparity among accuracy, sensitivity, and specificity, indicating that the data were not overfitted to the training set. We concluded that variability in the training data tends to reduce overfitting and overtraining as the deep neural network models learn to focus on the common pathologies. Future studies should demonstrate th...

show abstract

Section: Introductionmentioning

confidence: 99%

Feasibility of Deep Learning Algorithms for Reporting in Routine Spine Magnetic Resonance Imaging

Lewandrowski

Muraleedharan²,

Eddy³

et al. 2020

Int J Spine Surg

View full text Add to dashboard Cite

show abstract

“…This has, and will, lead to new situations where the developers of such intelligent machines are not able to fully predict their machines behavior and thus mistakes. The current study contributes to the existing literature (Malle et al, 2015;Awad et al, 2020;Bennett et al, 2020;Hidalgo et al, 2021) seeking to understand how situations of failure shape the public's attitude toward machines. Exploring how this is likely to unfold is a crucial step forward toward realizing the potential benefit of this technology.…”

Section: Discussionmentioning

confidence: 90%

“…Evidence suggests that people require AVs to be multiple orders of magnitude safer than human drivers (Liu et al, 2019). As argued in (Awad et al, 2020), negative public reaction may result in inflated prices of this technology (Geistfeld, 2017) and may shape how a tort-based regulatory scheme would turn out, both of which can influence the rate of adoption.…”

Section: Introductionmentioning

confidence: 99%

“…Understanding how we react to mistakes by machines (as compared to those by humans) is not an easy task. There is strong evidence that people react differently to mistakes made by machines and humans (Dietvorst et al, 2014(Dietvorst et al, , 2015Malle et al, 2015;Awad et al, 2020). There are also reasons to believe that people assign blame differently based on the difficulty of encountered situations.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Blaming Automated Vehicles in Difficult Situations

2020

Self Cite

View full text Add to dashboard Cite

Automated vehicles (AVs) have made huge strides toward large-scale deployment. Despite this progress, AVs continue to make mistakes, some resulting in death. Although some mistakes are avoidable, others are hard to avoid even by highly skilled drivers. As these mistakes continue to shape attitudes toward AVs, we need to understand whether people differentiate between them. We ask the following two questions. When an AV makes a mistake, does the perceived difficulty or novelty of the situation predict blame attributed to it? How does that blame attribution compare to a human driving a car? Through two studies, we find that the amount of blame people attribute to AVs and human drivers is sensitive to situation difficulty. However, while some situations could be more difficult for AVs and others for human drivers, people blamed AVs more, regardless. Our results provide novel insights in understanding psychological barriers influencing the public's view of AVs.

show abstract

“…Further, experiment results show that moral judgments on human drivers and AVs were similar ( Kallioinen et al, 2019 ). Consequently, many researchers emphasize the importance of including public morality and preference in AV ethics ( Awad et al, 2018b ; De Freitas et al, 2020a ; Savulescu et al, 2019 ; De Freitas et al, 2020a ; De Freitas et al, 2020a ). It is important to note that the focus of this study is limited to understanding acceptable AV moral behaviors for the public, which has been underexplored.…”

Section: Introductionmentioning

confidence: 99%

A Deeper Look at Autonomous Vehicle Ethics: An Integrative Ethical Decision-Making Framework to Explain Moral Pluralism

et al. 2021

View full text Add to dashboard Cite

The autonomous vehicle (AV) is one of the first commercialized AI-embedded robots to make autonomous decisions. Despite technological advancements, unavoidable AV accidents that result in life-and-death consequences cannot be completely eliminated. The emerging social concern of how an AV should make ethical decisions during unavoidable accidents is referred to as the moral dilemma of AV, which has promoted heated discussions among various stakeholders. However, there are research gaps in explainable AV ethical decision-making processes that predict how AVs’ moral behaviors are made that are acceptable from the AV users’ perspectives. This study addresses the key question: What factors affect ethical behavioral intentions in the AV moral dilemma? To answer this question, this study draws theories from multidisciplinary research fields to propose the “Integrative ethical decision-making framework for the AV moral dilemma.” The framework includes four interdependent ethical decision-making stages: AV moral dilemma issue framing, intuitive moral reasoning, rational moral reasoning, and ethical behavioral intention making. Further, the framework includes variables (e.g., perceived moral intensity, individual factors, and personal moral philosophies) that influence the ethical decision-making process. For instance, the framework explains that AV users from Eastern cultures will tend to endorse a situationist ethics position (high idealism and high relativism), which views that ethical decisions are relative to context, compared to AV users from Western cultures. This proposition is derived from the link between individual factors and personal moral philosophy. Moreover, the framework proposes a dual-process theory, which explains that both intuitive and rational moral reasoning are integral processes of ethical decision-making during the AV moral dilemma. Further, this framework describes that ethical behavioral intentions that lead to decisions in the AV moral dilemma are not fixed, but are based on how an individual perceives the seriousness of the situation, which is shaped by their personal moral philosophy. This framework provides a step-by-step explanation of how pluralistic ethical decision-making occurs, reducing the abstractness of AV moral reasoning processes.

show abstract

Drivers are blamed more than their automated cars when both make mistakes

Cited by 95 publications

References 28 publications

Feasibility of Deep Learning Algorithms for Reporting in Routine Spine Magnetic Resonance Imaging

Feasibility of Deep Learning Algorithms for Reporting in Routine Spine Magnetic Resonance Imaging

Blaming Automated Vehicles in Difficult Situations

A Deeper Look at Autonomous Vehicle Ethics: An Integrative Ethical Decision-Making Framework to Explain Moral Pluralism

Contact Info

Product

Resources

About