Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AccessShare: Co-designing Data Access and Sharing with Blind People

Rie Kamikubo, Farnaz Zamiri Zeraati, Kyungjun Lee, Hernisa Kacorri

TL;DR

AccessShare addresses the lack of accessible data inspection and control for blind contributors in AI datasets. It introduces a design probe and a co-design study with 10 blind participants to investigate how interactive consent and descriptor-based data inspection affect sharing decisions. The study finds that interactive dialogue and accessible descriptors facilitate communication with data stewards and reveal tensions between independence and interdependence in decision making, informing design recommendations for future accessible data-sharing platforms. Overall, the work demonstrates the value of participatory approaches in shaping inclusive AI data practices and highlights opportunities to improve descriptors, multiuser access, and consent mechanisms.

Abstract

Blind people are often called to contribute image data to datasets for AI innovation with the hope for future accessibility and inclusion. Yet, the visual inspection of the contributed images is inaccessible. To this day, we lack mechanisms for data inspection and control that are accessible to the blind community. To address this gap, we engage 10 blind participants in a scenario where they wear smartglasses and collect image data using an AI-infused application in their homes. We also engineer a design probe, a novel data access interface called AccessShare, and conduct a co-design study to discuss participants' needs, preferences, and ideas on consent, data inspection, and control. Our findings reveal the impact of interactive informed consent and the complementary role of data inspection systems such as AccessShare in facilitating communication between data stewards and blind data contributors. We discuss how key insights can guide future informed consent and data control to promote inclusive and responsible data practices in AI.

AccessShare: Co-designing Data Access and Sharing with Blind People

TL;DR

AccessShare addresses the lack of accessible data inspection and control for blind contributors in AI datasets. It introduces a design probe and a co-design study with 10 blind participants to investigate how interactive consent and descriptor-based data inspection affect sharing decisions. The study finds that interactive dialogue and accessible descriptors facilitate communication with data stewards and reveal tensions between independence and interdependence in decision making, informing design recommendations for future accessible data-sharing platforms. Overall, the work demonstrates the value of participatory approaches in shaping inclusive AI data practices and highlights opportunities to improve descriptors, multiuser access, and consent mechanisms.

Abstract

Blind people are often called to contribute image data to datasets for AI innovation with the hope for future accessibility and inclusion. Yet, the visual inspection of the contributed images is inaccessible. To this day, we lack mechanisms for data inspection and control that are accessible to the blind community. To address this gap, we engage 10 blind participants in a scenario where they wear smartglasses and collect image data using an AI-infused application in their homes. We also engineer a design probe, a novel data access interface called AccessShare, and conduct a co-design study to discuss participants' needs, preferences, and ideas on consent, data inspection, and control. Our findings reveal the impact of interactive informed consent and the complementary role of data inspection systems such as AccessShare in facilitating communication between data stewards and blind data contributors. We discuss how key insights can guide future informed consent and data control to promote inclusive and responsible data practices in AI.
Paper Structure (31 sections, 4 figures, 2 tables)

This paper contains 31 sections, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Prospects for Data Control in AI
  4. Participation in Data Stewardship
  5. Blind Contributors in Accessibility Datasets
  6. AccessShare: Our design probe
  7. Automatically Generated Descriptors
  8. Interface Design and Implementation
  9. Methods
  10. Recruitment and Participants
  11. Data Collection in Evaluation
  12. Data Access and Sharing in Co-design
  13. Part 1: Hands-on interface exploration.
  14. Part 2: Discussion with guiding questions.
  15. Analysis
  16. ...and 16 more sections

Figures (4)

  • Figure 1: A screenshot of AccessShare, the design probe in our study. Participants review their photos and labels in an HTML interface that contains a navigation menu along a page that provides a summary and a grid view of the photos with alt text based on descriptors for the selected data.
  • Figure 2: Findings in this paper come from a co-design session within a larger study that includes a background questionnaire, a system evaluation session with data capture, and a semi-structured interview on data sharing. Participants could indicate their data sharing decisions at any time after the interview once receiving an email with a unique link to their data in AccessShare.
  • Figure 3: Percentage of PII type across participants' photos.
  • Figure 4: When and how decisions were communicated.