By D. Maan, Jadetimes News
AI Empowering Independence for the Visually Impaired
When Matthew Sherwood shops for clothes, he often needs assistance to confirm the color or style he desires. Blind for over 15 years, Sherwood manages a family, a successful investing career, and has a guide dog named Chris. Despite his accomplishments, daily tasks like shopping can still challenge his independence.
The Role of AI in Everyday Tasks
Artificial intelligence is poised to offer significant help. Currently, Sherwood uses an app called Be My Eyes, which connects visually impaired users with sighted volunteers via live video to assist with tasks like matching clothes or checking expiration dates on milk. However, advancements in AI technology are starting to eliminate the need for human volunteers. Last year, Be My Eyes partnered with OpenAI to enable its AI model to see and describe what’s in front of users, showcasing the potential of AI to further aid visually impaired individuals.
AI Powered Assistive Technologies
In OpenAI’s latest demo, an AI powered version of Be My Eyes helped a user hail a taxi by instructing them when to raise their arm. Similarly, Google announced in May a feature for its "Lookout" app, designed to assist visually impaired users. These applications are part of a broader trend where AI is enhancing assistive technology, offering tools like eye tracking for controlling iPhones and detailed voice guidance for Google Maps users. Major tech companies, including Apple and Google, are expanding their AI powered tools to support people with various disabilities.
Transformative Potential of AI for the Disabled
Since the launch of ChatGPT, it has become evident that AI will transform numerous aspects of our lives, from work to communication. For people with disabilities, AI has the potential to be profoundly life changing. Sherwood highlights that previously, blind business professionals needed an assistant to read to them, but now AI provides them with unprecedented independence and opportunities. This technology enables blind individuals to gain employment, compete in business, and achieve success, marking a significant step forward in accessibility and empowerment.
Advancements in AI for Accessibility
Tech companies have been using early forms of AI to enhance product accessibility for years, such as automated closed captioning and screen readers. However, experts note that recent advancements in AI, driven by vast data sets and powerful computing systems, are accelerating possibilities in the assistive tech space. For instance, an AI tool that helps blind individuals hail taxis must be highly proficient at recognizing taxis, which necessitates extensive training on numerous examples.
Enhancing AI Tools with Generative Technology
A notable example of AI advancement is a Google tool designed for blind or low vision users, which describes what’s on their screen. This tool has been upgraded with a "question and answer" feature using Google's generative AI technology. Eve Andersson, Google’s senior director of product inclusion, equity, and accessibility, highlights that while AI's promise has been evident for years, it must reach a high quality level to be viable in products. Generative AI tools are particularly promising for accessibility as they can understand and produce information in various formats, including text, audio, photos, and videos.
AI Bridging Information Modalities
Generative AI’s ability to translate between different modalities makes it a powerful tool for accessibility. For instance, AI can convert audio into written text for hearing impaired users, acting as an intermediary to present information in the preferred medium. This versatility addresses the varied accessibility needs of individuals, whether related to hearing, vision, motor skills, speech, or cognitive functions.
Meeting Diverse Accessibility Needs
Eve Andersson emphasizes that accessibility needs vary widely but often revolve around how individuals perceive information. Disabilities can affect hearing, vision, motor skills, speech, and cognition, each requiring different modalities of information. AI’s ability to translate between these modalities is particularly valuable, making it an effective tool for enhancing accessibility and inclusivity in technology products.
Ongoing Investment in AI for Accessibility
Ensuring AI systems serve all users requires continuous investment. Because AI models are trained on human created data, they may replicate human biases, as seen in early instances where AI image generators struggled with race or algorithms displayed job ads based on gender stereotypes. To address this, major tech companies like Apple, Google, and Microsoft have partnered with the University of Illinois Urbana Champaign to create a diverse training dataset for AI speech recognition tools, including speech patterns from individuals with conditions like Parkinson's, Down Syndrome, and ALS.
Speech Accessibility Project
The Speech Accessibility Project collects recordings from volunteers with speech affecting disabilities. With over 200,000 recordings, a sample speech recognition tool created by the researchers now misunderstands speech only 12% of the time, down from 20% prior to training on the new dataset. Clarion Mendes, a speech language pathologist and clinical assistant professor, highlights the project's impact, noting that more diverse speech inputs improve AI’s ability to understand individuals who don't have "audiobook narrator" speech. This enhances independence and participation for those facing communication barriers.
The Business and Ethical Case for AI Accessibility
Investing in AI for accessibility is both ethically right and financially beneficial. Eve Andersson from Google emphasizes that inclusive technology can level the playing field and expand market opportunities by making products more appealing to government entities and educational institutions. By addressing diverse needs, AI technology not only supports individuals in their daily lives but also creates broader business advantages.
