Skip to main navigation Skip to search Skip to main content

A comparison of head-worn versus handheld artificial intelligence implementations for people with vision loss

  • Lighthouse Guild
  • New York University
  • Vrije Universiteit Amsterdam
  • Amsterdam UMC

Research output: Contribution to journalArticleAcademicpeer-review

7 Downloads (Pure)

Abstract

Significance: – The predominantly nonsignificant differences we found between head-worn (ARx) and handheld Seeing AI (artificial intelligence) implementations provide objective evidence to the ongoing debate about the relative advantages and disadvantages of form. People with vision loss must choose between these implementations based on functional needs and app accessibility, rather than form factor. Purpose: – To compare the functionality of AI implementations in head-worn devices versus handheld smartphones by objectively quantifying performance, usability, and acceptability when acquiring information from text and in daily activities. Methods: – A cross-sectional, counterbalanced, crossover design was employed to assess performance using Seeing AI in two formats (ARx headset and on a smartphone) and to compare two head-worn formats that utilize different AI algorithms—Seeing AI and Meta AI. Completion and timing were quantified for items in two task categories: Text and Searching & Identifying. Usability was evaluated with the System Usability Scale. Data were compared with a baseline condition with no assistive technology, and performance among AI implementations was assessed. Results: – There was no significant difference in the number of participants who completed tasks and timing between head-worn ARx and smartphone-based Seeing AI implementations. A comparison of two AI algorithms (Seeing AI and Meta) in wearable implementations found equivalent gains in performance but significantly faster task completion times for the Meta glasses. The timing advantage of Meta derives from its ability to provide more information about most tasks more quickly, whereas Seeing AI often requires additional prompts to gather sufficient data to complete tasks. The reported acceptability and usability were statistically similar among the three AI implementations. Conclusions: – We found no evidence to demonstrate an advantage in completing tasks using either head-worn (ARx and Meta glasses) or smartphone AI implementations.
Original languageEnglish
JournalOptometry and vision science
VolumePublish Ahead of Print
DOIs
Publication statusPublished - 12 Oct 2025

Fingerprint

Dive into the research topics of 'A comparison of head-worn versus handheld artificial intelligence implementations for people with vision loss'. Together they form a unique fingerprint.

Cite this