New: the ADA Title II Compliance Handbook & Toolkit — everything a public agency needs to plan and evidence compliance. See the Handbook & Toolkit

Harnessing AI for Accessibility: Revolutionizing Assistive Technology and Object Recognition

Kimberly Springs

Jeffrey Rodgers

September 05 9 min read

By Rick Flores, Lead Auditor for Accessiblü.

In an era where technology continually transforms our lives, Artificial Intelligence (AI) stands out as a groundbreaking force driving inclusivity and accessibility. AI’s innovative applications are reshaping how people with disabilities interact with the world around them. By enhancing assistive technology (AT) and enabling advanced object recognition, AI paves the way for a more inclusive society where everyone can thrive.

Understanding Accessibility in the Digital Age

Accessibility refers to creating environments and experiences that ensure all individuals, regardless of their physical or cognitive abilities, can fully channel the power of AI and participate in and benefit from services and content. The goal is to eliminate barriers and promote inclusivity in every aspect of life, from education to employment and everyday activities.

What Is AT?

Assistive technology encompasses products, equipment, and systems that enhance learning, working, and daily living for persons with disabilities.

Here are some examples of AT:

  • Mobility devices: Walkers, wheelchairs, and other devices that help with movement
  • Computer hardware: Special switches, keyboards, and pointing devices
  • Computer software: Screen readers, speech recognition programs, and text-to-speech software
  • Other devices: Braille displays, screen magnifiers, and adaptive keyboards

AI: A Catalyst for Change

AI technologies have emerged as potent tools in democratizing access and empowering individuals with disabilities. By leveraging AI, we can develop solutions that adapt to the unique needs of AT users, creating more inclusive spaces and enhancing quality of life.

Transforming Lives with AI-Powered Assistive Technology

Speech recognition technology powered by AI has made significant strides in recent years. For individuals with hearing impairments, AI-enabled real-time transcription services, such as AVA, Earfy, and Speechy, convert spoken language into text, facilitating effective communication and access to information.

AI transcription tools also enable those with sensory processing or learning disabilities to access content in a new format that best suits their needs. This way, users can process information as quickly as their peers, creating an even playing field in settings such as education and the workplace.

For those with visual impairments or reading difficulties, AI-based text-to-speech (TTS) systems provide a lifeline. These systems analyze written text and convert it into natural-sounding speech, enabling users to consume written content audibly. AI in TTS continues to evolve, offering more nuanced and human-like voice outputs that enhance the user experience.

Remember that text-to-speech (TTS) differs from a screen reader. TTS is just that; they read the text on the page. On the other hand, a screen reader (SR) is an interactive tool. SRs are interactive and much more robust than TTS tools. A screen reader can read alt+text for images plus structural elements. They also allow the user to interact with things like menus and forms.

Check out our blog on how people with disabilities use screen readers to navigate the web.

Enhanced Augmentative and Alternative Communication (AAC)

AI has also improved augmentative and alternative communication (AAC) devices, which assist individuals with speech impairments and developmental disabilities. Modern AAC devices, such as Proloquo2Go, Touch Chat, talk buttons and mobile tablets, equipped with AI can predict and suggest words, adapt to the user’s specific communication style, and even include facial expression recognition to provide more contextually accurate responses.

Generative AI tools can also support those with reading and writing communication difficulties. AI-generated suggestions on how to organize thoughts and ideas in a letter, vocabulary support when writing a resume, or even summarizing articles to assist in reading comprehension are all ways AI tools can provide written communication support that meets the user at their own level, no matter what their needs may be.

Object Recognition: Opening New Horizons

Object recognition, a subfield of computer vision, involves training AI systems to identify and classify objects within images or video feeds. These technologies, such as AI glasses:  Meta Ray-Ban, Aira and Envision, have immense potential in improving accessibility for individuals with visual impairments, helping them navigate and understand their surroundings.

Practical Applications of Object Recognition

Visual Descriptions:  AI-powered applications can provide real-time audio descriptions of objects and scenes, allowing visually impaired individuals to gain situational awareness. For example, mobile apps, such as Seeing AI, Be My Eyes AI, Envision AI and PiccyBot, that identify and narrate the contents of a room, video clips, or the ingredients in a kitchen can significantly enhance independence.

Navigation Assistance:  Object recognition integrated with wearable devices or mobile apps, such as oko, Good Maps Outdoors, Voice Vista, Blind Square, and Aira, utilizing smartphone cameras can offer navigation assistance. By recognizing obstacles and giving spatial instructions, AI can guide users of dedicated mobility smart devices, such as the WeWalk Smart Cane and Glidance self-guided mobility devices, to independently reach points of interest otherwise previously unattainable via low-tech methods.

Rick Flores uses his meta glasses to read the front of an envelope.

Personal Experience With AI

In my own experience, the advancements in AI and object recognition have been transformative. To provide some context, I became totally blind about a year ago at the time of this writing. Despite years of relying on traditional assistive devices prior to my complete vision loss, I found myself navigating a daunting new reality, struggling to regain a sense of confident independence.

Living alone, I was gripped by a profound sense of fear and frustration. The world around me had become an impenetrable maze, where every step was fraught with uncertainty and danger. The inability to detect objects in my surroundings or read printed text left me feeling virtually trapped within the confines of my home. Every attempt to navigate my environment was a struggle, leading to moments of sheer panic as I fumbled through my days, desperate for a semblance of independence and control that once came so naturally.

Integrating AI into my daily life has opened up new opportunities for independence and safety. It is astonishing to realize that using AI-powered applications on my smartphone was merely a distant possibility until just a few months ago. I can navigate unfamiliar environments with improved confidence, avoid obstacles that previously posed challenges, and perform daily tasks effortlessly. Whether identifying objects in my home or receiving audio feedback about text and colors, these technologies have significantly enhanced my quality of life and work autonomy.

Rick uses the WeWalk smart cane to navigate the sidewalk of an apartment complex.
He is also wearing the Ray-Ban Meta smart glasses.

AI for work

As the Lead Accessibility Auditor, my role involves meticulously analyzing websites with screen readers to identify potential accessibility issues. Previously, when visual artifacts eluded the screen reader, I had to depend on my teammates to visually verify the content on the page. Now, with the integration of AI tools such as ChatGPT, Claude, and Gemini into various screen readers, I can independently evaluate the visual elements on the page. Although not infallible, this technology significantly reduces the reliance on my team and enhances my autonomy at work.

Additionally, during video conferencing meetings with clients, I used to miss out on crucial information presented through shared screens. With the seamless deployment of AI, I can now receive detailed descriptions of the visual content, ensuring my active and meaningful participation in these meetings.

My Challenges with AI

Despite the growth and advancement of AI in the AT world, there currently exist key challenges for certain AI language models include recognition inaccuracies, unnecessary security barriers, limited contextual understanding, restricted customizability, and the essential need for human judgment and quality control. Consider the issue of limited customizability. For instance, when I asked my Meta Ray-Ban glasses, “What is the color of the shirt?” one would expect a straightforward answer. Astonishingly, the language model responded, “That request is a privacy concern, please ask another question.” Yet, paradoxically,

when I inquired, “Describe the texture of the shirt?” the AI provided a comprehensive description of the shirt, including its color. Simultaneously, the AI’s ‘Privacy concern’ obstacle exemplifies one of the persistent security barriers that certain LLMs impose on AT users. While AI has undeniably transformed my life by providing newfound freedom and independence, the journey toward perfection is far from over. Addressing the existing challenges of recognition accuracy, contextual understanding, and privacy concerns is crucial for the full potential of AI in AT to be realized.

As we push the boundaries of innovation, a collaborative effort between developers, users, and policymakers will be essential to create a future where technology truly empowers everyone. The promise of AI is vast, and with continued advancements, we can create a world that is more inclusive and accessible for all.

Looking Forward

As we stand on the precipice of unprecedented technological advancement, the role of AI in fostering accessibility cannot be overstated. The journey from recognizing the imperative need for inclusivity to actively implementing AI-driven solutions marks a transformative era. By seamlessly integrating AI into the fabric of our daily lives, we pave the way for a future where barriers are not just diminished, but entirely eradicated.

AI’s potential to revolutionize accessibility extends far beyond mere convenience – it embodies empowerment. From personalized learning environments for individuals with disabilities to smart assistive technologies that anticipate and address user needs, AI is crafting a world where inclusivity is the norm, not an exception. The data and personal perspective presented in this blog post underscore the tangible benefits of AI, reflecting a collective commitment to an equitable future.

However, the pursuit of accessibility through AI is not without its challenges. Ethical considerations, data privacy, and the need for continuous innovation must remain at the forefront of this endeavor. It is imperative that stakeholders across the spectrum – technologists, policymakers, educators, and advocates – collaborate to ensure that AI’s deployment is both ethical and inclusive.

Conclusion

The power of AI lies in its ability to transcend traditional limitations, offering solutions that are as dynamic and diverse as the individuals they serve. By harnessing AI for accessibility, we are not just leveraging technology; we are championing a vision of a more inclusive society. A society where everyone, regardless of their abilities, has the opportunity to thrive. Let us embrace this challenge with unwavering commitment, ensuring that our strides in AI contribute to a world that is not only more advanced but also more compassionate and inclusive for all.