AI Smart Eyeglasses Show Promise for People with Low Vision: JAMA
USA: Researchers have found in a new study that AI-powered smart eyeglasses may serve as a useful assistive tool for individuals with low or no vision.
The devices performed best in identifying common objects, recognizing neutral colors, and reading simple text such as children's books. However, current technological limitations may affect their overall usability. As artificial intelligence continues to evolve, these smart glasses are expected to become more capable, although additional research involving people with visual impairment is needed to better understand their real-world benefits and limitations.
The findings are from a pilot feasibility study published in JAMA Ophthalmology by Robert J. Medina from the Ocular Oncology Service at Wills Eye Hospital, Thomas Jefferson University, Philadelphia, and colleagues. The researchers assessed the performance of second-generation Ray-Ban Meta AI smart eyeglasses across a range of vision-related tasks, including object recognition, object description, reading, and money identification.
Artificial intelligence-enabled wearable devices are increasingly being explored as tools to support people with vision loss. Despite growing interest, evidence on their practical accuracy and reliability has remained limited. To address this gap, the investigators evaluated how effectively the AI-powered eyeglasses could identify and describe both single and multiple objects under standardized testing conditions.
The pilot case series included six fully sighted study investigators who evaluated the AI smart eyeglasses in a controlled setting. The researchers assessed the device's accuracy in object identification, color recognition, object direction, object counting, reading various types of text, and identifying and counting US currency.
The study led to the following findings:
- AI smart eyeglasses identified common objects with 99% accuracy.
- Paper currency was correctly identified with 91% accuracy.
- The devices achieved 93% accuracy when reading children's books.
- Handwriting recognition was accurate in 88% of cases.
- Object direction or orientation was identified with 83% accuracy.
- Color discrimination accuracy was 64%.
- Standard printed text was read correctly with 59% accuracy.
- Object counting accuracy was 50%, indicating difficulty with counting multiple objects.
- Coin identification performed poorly, with only 2% accuracy.
The researchers acknowledged that the study was limited by its small sample size, controlled laboratory setting, restricted range of test objects, and inclusion of only fully sighted study investigators, which may limit the generalizability of the findings.
The authors concluded that AI smart eyeglasses show promise as an assistive tool for people with low or no vision, particularly for object recognition, reading simple text, and identifying paper currency. However, improvements in color discrimination, object counting, and coin recognition are needed. They called for larger studies involving individuals with visual impairment to better evaluate the technology's real-world effectiveness.
Reference:
Medina RJ, Botros S, Gandhi P, et al. Artificial Intelligence Smart Eyeglasses for the Detection and Description of Stationary Objects. JAMA Ophthalmol. Published online June 25, 2026. doi:10.1001/jamaophthalmol.2026.2185
Next Story