AI & Accessibility

In recent years, the words “artificial intelligence” (AI) have worked their way into almost every aspect of daily life, marking a turning point in the way we will use technology for years to come. But outside of ChatGPT, Gemini, and robot food delivery, AI is making exciting, tangible advancements in adaptive products and accessibility for the disabled community. 

AI has made big improvements in assistive technology for blind individuals and those with low vision. For example, a screen reader is software that can read aloud to the user everything occurring on screen. While this technology has been around since the 1980s, AI can provide a more natural voice, describe images without alternative text, and summarize documents that haven’t been properly tagged for accessibility. These new enhancements provide an opportunity to enjoy content that a traditional screen reader might have skipped over. This could make more of the internet, such as older or international websites, accessible to those who are blind or have low vision. Flagship smartphones from Samsung, Google, and Apple are all increasingly incorporating smart screen readers such as this, and desktop screen readers are incorporating AI software, such as JAWS’ new PictureSmart AI. 

The white cane, a widely used mobility aid for blind individuals, is now receiving AI-powered upgrades. Several products are entering the market, including the WeWALK Smart Cane™. This device uses AI to detect overhead obstacles, provides a voice assistant for travel directions (via subscription), identifies nearby businesses and landmarks, and connects to a user’s smartphone, enabling hands-free navigation. While not necessarily a degree of involvement every user would want, AI is creating new choices in the lifestyles of those who use white canes to navigate.

Another field in which AI is making significant advancements is communication. Speech-to-text technology can convert a person’s speech into written text, opening new doors for people who are not able to use a keyboard, and speech recognition is made much stronger with AI technology. On the flipside, text-to-speech technology supports blind users and users with low vision, as well as individuals with cognitive or reading disabilities to receive information through written word being spoken aloud. AI allows this generated speech to sound more natural and even convey emotion. In some cases, people can even create custom voices. For example, the company The Voice Keeper Kids uses AI technology to allow families to create a custom voice for their non-speaking child using recordings of a friend or loved one’s voice.  

However, text-to-speech alone will not satisfy everyone’s communication needs, even if its tone is more natural. For those who use sign language, newer advancements have appeared, including apps such as Slait, an AI program which can convert both sign language and speech into text, helping to facilitate real time conversation on video calls, and the program Signapse.ai, which can translate written text into American Sign Language (ASL) or British Sign Language (BSL), posing as an alternative or supplement to subtitle-first accommodations. The program Sign-Speak is the first to translate ASL to spoken English and back again, breaking new ground for people who communicate with, or ingest information through, sign language even in cases where a human sign language interpreter is not available. Improved speech recognition can make these products easier to use, and allows for more accurate translation in programs like Sign-Speak. Broadening communication abilities can create a safer and more open environment, and if used correctly, artificial intelligence could be a significant step in this progression.

For those with amputations, AI-powered prosthetic limbs can significantly enhance independence and functionality. AI algorithms and machine learning allow prosthetics to better mimic human motion and improve coordination. For example, a smart prosthetic hand can determine the texture of an object being picked up and adjust its grip accordingly, while also giving feedback based on objects’ temperature and pressure.

Another useful advancement for those with mobility disabilities is that of voice and gesture commands. With the aforementioned improvements in speech recognition, it will become easier and more natural to control devices in one’s environments with voice commands. Similarly, machine learning has opened the door for improved gesture command recognition. With technology like this, a person could control much of their home without needing to physically interact with switches or controls. 

Of course, alongside these exciting technological advancements come ethical questions and issues. AI training data, the collections of information AI models use to learn patterns, often exclude disabled individuals’ experiences or treat them as statistical outliers. This can lead to models that fail to recognize diverse speech patterns, body movements, or assistive device interactions. In the sign language interpreting programs discussed earlier, there is the complication of the language’s nuances. Cultural and regional phrases, as well as slang, could be misinterpreted by an AI algorithm lacking that critical cultural context. 

Additionally, in-home AI technology that relies upon constantly recording our environments could come with a privacy risk, as audio recorders, cameras, or movement sensors could be at risk for hacking. While this invasion of privacy would certainly be protected against, there is still some risk and a natural discomfort with having WiFi-enabled audio and video recording equipment in one’s home.

Finally, there is the age-old problem of expense, most frustrating when it comes to products that are meant to increase accessibility. In the current system, it is unfortunately true that not every disabled person has the same access to helpful resources, especially expensive cutting-edge technology like the WeWALK Cane or fully AI-enabled smart homes. This unfair distribution of helpful technology is called an access divide. So even if these new advancements are exciting, they may not impact the average person with a disability for many years, despite how much they could change that person’s life.

AI technology’s impact on the disabled community could be very exciting, but it comes with the same risks and issues as any other technological advancements. In this time of change, it becomes more and more evident that disabled individuals should be heavily involved in incorporating this new technology into our society, a concept known as participatory design. By allowing disabled people to be part of designing, testing, and distributing the technology that will aid their communities, we can be more assured that the technology serves those it is meant to in the right way. While this could help solve the bias and privacy risks, the issue of access divide remains. Unless disability aids and assistive technology are valued more as the necessity they are, and making them better is seen as vitally important rather than a luxury, they might not be able to help all the people they could. It’s important for our policymakers, developers, and our culture as a whole to see accessibility not as an afterthought, but as a foundational principle: ensuring that AI empowers everyone, not just the privileged few.

About VisitAble

VisitAble is a disability training and consulting company dedicated to making accessibility more intuitive and inclusive. We focus on helping organizations support individuals across all types of disabilities. Disabilities affect children, young adults, veterans, first responders, seniors, older adults, parents, and more. There are many visible disabilities, but there are also many invisible disabilities. Our goal is to create environments where everyone belongs, with tools and resources to empower organizations to welcome everyone.