Three CSE papers at CHI 2021 recognized with honorable mentions

The papers dealt with issues of accessibility, privacy, and consent in technology.

Girl with an iPad Enlarge

Three papers co-authored by U-M CSE researchers earned Honorable Mention at the 2021 Conference on Human Factors in Computing Systems (CHI), the premier international conference on Human-Computer Interaction. Seven students and faculty were involved in the works, which dealt with issues of accessibility, privacy, and consent in technology.

Learn more about the papers:

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

Cynthia L. Bennett (Carnegie Mellon University), Cole Gleason (Carnegie Mellon University), Morgan Klaus Scheuerman (University of Colorado Boulder), Jeffrey P. Bigham (Carnegie Mellon University), Anhong Guo (University of Michigan), Alexandra To (Northeastern University)

Image descriptions play an important role in increasing access to visual information online to people who are blind. Content creators are instructed to write these textual descriptions to make their content accessible, but existing guidelines lack specifics on how to write about people’s appearance, particularly while remaining mindful of consequences of (mis)representation. This paper reports on interviews with screen reader users who were also Black, Indigenous, People of Color, Non-binary, and/or Transgender on their current image description practices and preferences, and experiences negotiating theirs and others’ appearances non-visually. The researchers discuss these perspectives, and the ethics of humans and AI describing appearance characteristics that may convey the race, gender, and disabilities of those photographed. They also share considerations for more carefully describing appearance.

PrivacyMic: Utilizing Inaudible Frequencies for Privacy Preserving Daily Activity Recognition

Yasha Iravantchi (University of Michigan), Karan Ahuja (Carnegie Mellon University), Mayank Goel (Carnegie Mellon University), Chris Harrison (Carnegie Mellon University), Alanson Sample (University of Michigan)

Sound is invaluable to enabling activity recognition in computing systems, and microphones are what allow them to perform tasks as our digital assistants. However, the microphones we use daily are optimized for human speech and hearing ranges: capturing private content, such as speech, while omitting useful, inaudible information that can aid in acoustic recognition tasks. In this paper, the researchers simulated acoustic recognition tasks using sounds from 127 everyday household and workplace objects, finding that inaudible frequencies such as ultrasound and infrasound can act as a substitute for privacy-sensitive frequencies. To take advantage of these inaudible frequencies, they designed a Raspberry Pi-based device that captures inaudible acoustic frequencies with settings that can remove speech or all audible frequencies entirely. They conducted a perception study, where participants “eavesdropped” on PrivacyMic’s filtered audio, and found that none of the participants could transcribe the recorded speech. PrivacyMic was able to achieve over 95% classification accuracy in real-world activity recognition across all environments.

Yes: Affirmative Consent as a Theoretical Framework for Understanding and Imagining Social Platforms

Jane Im (University of Michigan), Jill Dimond (Sassafras Tech Collective), Melody Berton (Sassafras Tech Collective), Una Lee (And Also Too), Katherine Mustelier (University of Michigan), Mark Ackerman (University of Michigan), Eric Gilbert (University of Michigan)

Affirmative consent is the idea that someone must ask for—and earn—enthusiastic approval before interacting with another person, sometimes referred to by the shorthand “yes means yes.” In this paper, the researchers explore how this principle can be used to theorize online interactions, using feminist, legal, and HCI literature to analyze social computing systems. They argue that affirmative consent is theoretically useful for explaining problematic online phenomena like mass harassment, revenge porn, and problems with content feeds. They also argue that it can be a basis for new design ideas in consentful socio-technical systems.