ArticleTranslational vision science & technology2025
Using ChatGPT To Generate Walking Directions-Could It Potentially Help Visually Impaired With Micronavigation?
Article in Translational vision science & technology, 2025. The graph could read no effect estimate from its abstract, so it casts no vote on the map. Cited by 2 papers.
What it found
Each row is one number read from the abstract, on the scale the paper reported it, with its interval. Left of the dashed line favours the treatment, right favours the comparator. Under each row is the sentence it came from. New to these charts? A ten-minute tutorial.
The abstract states no effect estimate the extractor could read, or names no intervention and outcome on the map, so this paper lights no cell and moves no belief. It is still indexed, cited and linked below.
The trial behind it
Trials whose registry record cites this paper, or whose number appears in the abstract. A trial that started after this paper was published is citing it as background, not reporting it.
Neither the registry nor the abstract names a trial number. If this is a trial report, that itself is worth knowing.
Who cites it
2 citing papers in PubMed.
- Recent Developments in Research to Make the World Better for People with Low Vision.Ophthalmic & physiological optics : the journal of the British College of Ophthalmic Opticians (Optometrists) · 2026Review
- Article
Corrections and comments
PubMed lists nothing against this paper. Absence here is not a guarantee, only a check that was made.
Authors and funding
3 authors.
Funding
No grant is acknowledged in the PubMed record.
Abstract
Purpose: Micronavigation poses challenges for blind and visually impaired individuals (BVI). We investigated the potential of using ChatGPT4o to provide navigation directions. Methods: We created a dataset of real-world micronavigation scenarios consisting of 113 scene images and corresponding human-generated scene descriptions. A total of 407 human-generated wayfinding queries and their expected responses were used for evaluation (245 answerable and 162 not answerable based on the scenes). "I do not know" as a response was expected for the negative cases. Sensitivity (SEN) and specificity (SPE) of navigation directions output by ChatGPT4o was evaluated under different input conditions. Results: The default ChatGPT4o, with scene images as inputs, resulted in SEN and SPE values of 64% and 75.9%, respectively. Instructing via prompts on how to respond to unanswerable questions did not affect SEN, but increased SPE by 14 %. SEN and SPE both decreased by about 13% and 7%, respectively, when ChatGPT4o first generated text description from the input images and then answered the queries based the generated descriptions. When human-written scene descriptions were provided as input, SEN and SPE improved by about 17% and 16%, respectively. Conclusions: Some spatial information needed for micronavigation was diminished in generative scene descriptions. This problem may be addressed by one-step vision language models. General-purpose ChatGPT4o with fine-tuned prompts was still inferior to human in scene understanding. A multimodal large language model custom-trained on navigation scenarios might help improve the performance. Translational Relevance: AI chat-bots may have the potential to provide navigation assistance to BVI to a certain degree.
Indexed as
Identifiers
What OpenQuestion holds
Registered trials
Read under generation 80e0d062 · epoch 390. Bibliography from PubMed, PubMed Central and OpenAlex; grants from NIH RePORTER; trial links from ClinicalTrials.gov; estimates, votes and beliefs from the OpenQuestion graph.