Project Overview
Project Summary:
Broady is a mobile application that connects the visually impaired and sighted individuals through one-on-one chat services and artificial intelligence. Users needing visual assistance can send images and receive instant image descriptions via AI chatbot AI Rody and human interpreters. In particular, AI Rody uses the GPT-4V API, along with the Naver Clova OCR API and the Bing Visual Search API, to overcome language-specific issues and offer precise image descriptions in Korean. This solution aims to increase accessibility, empowering visually impaired users to navigate online and offline environments with greater autonomy.
Identifying the Challenge
The Social Problem:
Korea’s Anti-Discrimination Act requires that all content be accessible to the visually impaired. Yet over 96% of websites fail to meet accessibility standards. Particularly, the absence of alternative texts, limits the visually impaired from understanding visual content, making it inevitable for them to rely on others for assistance. Increasing accessibility is crucial to promote inclusivity for all in the digital sphere.
Innovation and Uniqueness
Why Our Project Stands Out:
Broady sets itself apart from its competitors due to its innovative blend of AI and human interpreters. Unlike existing solutions that rely solely on either AI or human support, Broady uses them both, offering instant yet precise visual assistance. Furthermore, Broady's visual data will be utilized to create alternative texts for businesses, thus addressing web accessibility issues at a fundamental level.
Insights and Development
Learning Journey:
After launching an MVP that didn't meet our user engagement expectations, our team realized the importance of aligning our development process with user's real needs. Simplifying the user flow, we also actively incorporated feedback from visually impaired individuals and welfare centers. We even established an advisory board of blind people to address inconveniences. This user-centered change significantly improved Broady’s recognition.
Development Process:
We initially conducted an MVP to provide image commentary via KakaoTalk. This trial highlighted immediacy, specificity, and uniqueness as components of effective commentary. Our team identified the shortcomings of existing AI photo commentary services, and developed Broady, a service that combines image analysis AI and human input. Broady's interactive UI enhances accessibility for visually impaired users.
