"Siri Talks at You": An Empirical Investigation of Voice-Activated Personal Assistant (VAPA) Usage by Individuals Who Are Blind
Ali Abdolrahmani, Ravi Kuber, Stacy M. Branham · 2018 · Proceedings of the 20th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2018) · doi:10.1145/3234695.3236344
Summary
This UMBC and UC Irvine study is one of the first to examine how individuals who are blind use voice-activated personal assistants (VAPAs) like Siri, Amazon Echo, and Google Home, and what barriers they encounter. The researchers interviewed 14 legally-blind adults (ages 21-66, mean 31) who used both mobile and home-based VAPAs. All participants relied on screen readers as their primary access tool and used VAPAs on a daily or weekly basis. While sighted users tend to view VAPAs as "entertaining" or "gimmicky," blind participants treated them as serious accessibility tools — VAPAs enabled access to otherwise inaccessible technologies like touchscreen thermostats, microwave ovens, and light switches, and reduced the time and effort required for tasks compared to touchscreen interaction with VoiceOver. The study covered three areas: interaction experiences, accessibility of VAPAs, and perceptions of privacy and security. Participants used VAPAs across six task categories: managing time (reminders, alarms, calendar), information seeking (directions, weather, facts), accessing apps/services/devices (phone calls, smart home control, Uber), media and entertainment, math and language aid, and accessing third-party apps.
Key findings
Participants identified multiple usability challenges specific to their blindness. Input challenges included misinterpretation of commands in noisy environments (10 participants), frustrating timeouts when composing complex commands like calendar entries, difficulty with accent/colloquial phrase recognition, and Siri misinterpreting contact names repeatedly without learning. Output challenges were significant: responses were often too verbose (e.g., receiving a full restaurant address when only asking for opening times) or too brief, with no ability to customize the level of detail. Experienced screen reader users found VAPA speech too slow but could not adjust the rate — five participants noted that faster output would also improve privacy by making speech unintelligible to bystanders. Visual cues had no non-visual alternatives: Amazon Echo's light ring blinks for notifications and changes color for muting, but blind users could not detect these states, creating an information asymmetry with sighted users in shared spaces. Eight participants compared platforms, with a consensus that Siri "talks at you" while Echo and Google Assistant felt more conversational and better handled follow-up queries. Privacy and social concerns strongly shaped usage patterns: 7 of 14 avoided using VAPAs on public transport, 9 described social awkwardness, and one participant recalled Siri loudly announcing "brush your teeth" at school. At home, VAPAs were a "comfort zone" — some participants considered Echo a "perfect fit" and "inseparable" part of daily life. Participants expressed concern that the trend toward adding visual displays to VAPAs (like Amazon Echo Show) would create a gap between blind and sighted usage of these inherently voice-first platforms.
Relevance
This study provides critical design guidance for making voice assistants truly inclusive. The central finding — that VAPAs are "serious tools" for blind users, not entertainment — has direct implications for how voice AI companies prioritize development. Several specific design recommendations are immediately actionable: provide customizable speech rate and verbosity levels per-task, create non-visual alternatives to visual status indicators (earcons for the Echo light ring), extend command timeouts for complex inputs, add "discreet" or "whisper" modes that mask personal details in public, and ensure voice-only interaction pathways remain fully functional as visual displays are added. The finding that sighted users can recover from recognition errors visually (scanning the screen for the intended result) while blind users cannot highlights a fundamental asymmetry — errors disproportionately impact blind users, making speech recognition accuracy and graceful error recovery especially critical for this population. The concern about VAPAs evolving toward visual interfaces is prescient and remains highly relevant as smart speakers increasingly incorporate screens, potentially eroding the accessibility gains that voice-first interaction provided.
Tags: blindness · voice interface · voice assistant · Siri · Amazon Echo · Google Home · smart speaker · mobile accessibility · privacy · social acceptability