“It’s Complicated”: Negotiating Accessibility and (Mis)Representation in Image Descriptions of Race, Gender, and Disability

Bennett, Cynthia L.; Gleason, Cole; Scheuerman, Morgan Klaus; Bigham, Jeffrey P.; Guo, Anhong; To, Alexandra

doi:10.1145/3411764.3445498

Cited by 73 publications

(36 citation statements)

References 74 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Prior work has studied VAT usage, including the types of images and questions blind people share with VATs [19,25,44,46,47], what information blind people want in image descriptions [40ś43, 95], ethical considerations regarding AI [13], user-centered evaluations of the accuracy of AI-powered visual descriptions [70], as well as how remote sighted assistants provide visual assistance [64,90].…”

Section: Blind People's Use Of Vats and Privacy Concernsmentioning

confidence: 99%

“…To identify updates to the policies, we irst observed if there was a change in the publication date, then compared the initial 2019 dataset to the 2021 policies on the VAT companies websites using the Track Changes->Compare Documents tool in Microsoft Word. Eight of the 13 companies had updated their policies between December 2019 and May 2021 13 . We observed only one change that corresponded to our annotation protocol: Amazon's privacy policy now directly speciies that video and image data are collected, e.g., łWhen you use our voice, image and camera services, we use your voice input, images, videos, and other personal information to respond to your requests, provide the requested service to you, and improve our services.…”

Section: Policy Validationmentioning

confidence: 99%

See 1 more Smart Citation

Privacy Concerns for Visual Assistance Technologies

Stangl

Davis

Xie

et al. 2022

ACM Trans. Access. Comput.

View full text Add to dashboard Cite

show abstract

Section: Blind People's Use Of Vats and Privacy Concernsmentioning

confidence: 99%

Section: Policy Validationmentioning

confidence: 99%

Privacy Concerns for Visual Assistance Technologies

Stangl

Davis

Xie

et al. 2022

ACM Trans. Access. Comput.

View full text Add to dashboard Cite

show abstract

“…The problem of inclusion in technology design is, to borrow language from HCI scholar Cynthia Bennett, "slippery" at best [9]. On one hand, training Alexa on queer voices from the queer bodies of queer people might, indeed, result in a voice-recognition algorithm that can more adequately distinguish between queer voices.…”

Section: Queering Sonic Technologiesmentioning

confidence: 99%

Sonic Technologies of a Queer Breakup

Kinnee

Rosner

Desjardins

2022

Designing Interactive Systems Conference

View full text Add to dashboard Cite

Over the past decade, the growth of voice assistants has presented new challenges within domestic life. Prior research has shown that such technologies afect users unevenly or fail to account for some relationships and domesticities entirely. Our work investigates the under-examined topic of queerness at home. Drawing on the experience of queer breakup, we describe a design inquiry and a frst-person research approach exploring two concurrent relationships in separate households both using Alexa. We explore issues of temporality, glitch, and shared accountability. We also ask critical questions with audio experiments, including: How do voice assistants diferentiate between queer voices? How should we converse with voice assistants about queerness? And are voice assistants "queer enough"? We contribute a discussion of diference, inclusion, and queer cultures of adversarial use to highlight the limitations of both everyday and professional language for describing and analyzing the particulars of queerness and interaction design. CCS CONCEPTS• Human-centered computing; • Interaction design; • Interaction design theory, concepts and paradigms;

show abstract

“…Like many works in the computer vision field, we rely on images with crowdsourced and inferred demographic labels. Both processes have been criticized for their lack of consent [41], the way they operationalize identity [81], and the harm they may cause through mis-identification [14]. These problems reinforce the need for highquality, consensual demographic data as a means to improve ethical norms and defend against adversarial ML attacks.…”

Section: Ethicsmentioning

confidence: 99%

Subverting Fair Image Search with Generative Adversarial Perturbations

Ghosh,

Jagielski,

Wilson

2022

Preprint

View full text Add to dashboard Cite

In this work we explore the intersection fairness and robustness in the context of ranking: when a ranking model has been calibrated to achieve some definition of fairness, is it possible for an external adversary to make the ranking model behave unfairly without having access to the model or training data? To investigate this question, we present a case study in which we develop and then attack a stateof-the-art, fairness-aware image search engine using images that have been maliciously modified using a Generative Adversarial Perturbation (GAP) model [75]. These perturbations attempt to cause the fair re-ranking algorithm to unfairly boost the rank of images containing people from an adversary-selected subpopulation.We present results from extensive experiments demonstrating that our attacks can successfully confer significant unfair advantage to people from the majority class relative to fairly-ranked baseline search results. We demonstrate that our attacks are robust across a number of variables, that they have close to zero impact on the relevance of search results, and that they succeed under a strict threat model. Our findings highlight the danger of deploying fair machine learning algorithms in-the-wild when (1) the data necessary to achieve fairness may be adversarially manipulated, and (2) the models themselves are not robust against attacks. CCS CONCEPTS• Information systems → Retrieval models and ranking; • Security and privacy;

show abstract

“It’s Complicated”: Negotiating Accessibility and (Mis)Representation in Image Descriptions of Race, Gender, and Disability

Cited by 73 publications

References 74 publications

Privacy Concerns for Visual Assistance Technologies

Privacy Concerns for Visual Assistance Technologies

Sonic Technologies of a Queer Breakup

Subverting Fair Image Search with Generative Adversarial Perturbations

Contact Info

Product

Resources

About