​Understanding what Assistive Technology is ​

As we consider the future of assistive technology (AT), we need to define it first. AT is any item (device, software, or tool) that helps someone with a disability or functional limitation increase their independence.

Different types of AT can assist a person in a variety of areas:  

• ​For a person with physical limitations, a specialized keyboard can make the task of typing easier.  
• ​For a person who is blind, a screen reader may allow them to use their computer.  
• ​For a person who is hard of hearing, captioning may allow them more inclusion in a meeting. 

​AT and AI 

​One area of AT that has seen great and exciting improvements is artificial intelligence (AI). The term “artificial intelligence” was introduced in 1955 but has become popular in the last decade due to deep learning breakthroughs. 

​Today we all benefit from, or actively use AI in our daily lives. We leveraged it to provide this article! If you stream movies through an online provider, have you wondered how it recommends movies for you?  Those tools use a form of AI called Machine Learning (ML). Have you asked virtual assistants (like smart home devices) where the nearest coffee shop is? Both use AI Natural Language Processing (NLP). Computer Vision is another example of AI that provides facial recognition on your phone, or can detect a tumor in an X-ray.  

​Positive Impacts for People with Disabilities 

​These are just a few examples of how we use AI. For many of us it is very convenient. For a person with a disability, AI can be life changing. Here are some examples of how this is currently happening. 

​John was recently diagnosed with ALS and over time he will lose his ability speak. As his disease progresses, it is important to John that his voice is still heard. He is currently working with his Speech Language Pathologist to “bank” his voice. Voice banking is a process that allows individuals to record and preserve their voice for future use. AI plays a central role in capturing, processing, and reproducing natural sounding, personalized speech. 

​Amir is legally blind and is grateful for the AT that he uses in his daily life. One tool that he uses is a pair of specialized AI-powered glasses. The eyewear provides real-time audio descriptions of visual information such as printed text on his mail or an email on his computer. Additionally, the glasses can tell him what his shirt looks like or let him know who walked into the room. With a wrist movement it will tell him the time of day. If he holds money in front of the glasses, they tell him the value of the bill. 

​Ellen is a professional who has dyslexia. Through the years she has struggled to write reports, draft emails, and take notes during meetings. Through the use of AI, today Ellen is able to perform these tasks without the frustration she previously faced. Built-in features in Microsoft 365 such as Microsoft Editor include spelling and grammar correction tools, and refinement tools for clarity, conciseness, punctuation conventions and vocabulary. Additionally, she can have the computer read the text aloud to her. While attending Microsoft Teams meetings she uses the transcribe feature, which captures everything spoken. This allows Ellen to focus on the content knowing that she can access the transcript later as a notes tool. 

​What’s next? 

​Self-driving vehicles! Minnesota has an exciting project called goMARTI – Minnesota’s Autonomous Rural Transit Initiative. This self-driving shuttle pilot program has three ADA-compliant vehicles that use data from cameras, radar, GPS, and LiDAR (Light Detection and Ranging remote sensing technology) to understand their current environment. Additionally, goMARTI can recognize and respond to pedestrians, vehicles, and other obstacles. The shuttle can make decisions regarding lane changes and turns without human intervention. 

​Another cutting-edge AI tool in development is the humanoid robot. Dr. Arshia Khan from the University of Minnesota Duluth has conducted a number of clinical trials at nursing homes throughout the U.S. using humanoid robots to enhance delivery of care in persons affected with Alzheimer’s and related dementia. The robots can facilitate social interactions, stimulate group activities, administer therapies, assist with physical and emotional wellbeing. Dr. Khan’s work demonstrates that humanoid robots can be help patients stay independent longer.  

​Challenges 

​These are just a few examples of AI’s current and potential impacts. It is important to remember even though AI can provide many benefits for people with disabilities, there are also potential challenges. 

​Accessibility barriers may exist. Some of the tech may have touchscreens, small fonts, or require mouse use that can prevent some users with motor, visual or cognitive disabilities. Additionally, algorithms may reinforce social biases which can make the tool less helpful or discriminatory. 

​Some AI systems make decisions or assumptions on behalf of users which can cause over-automation. For example, a navigation system may reroute someone with a mobility impairment based on an assumption about their ability. It may route them to avoid stairs where their personal preference may be a shorter route with assistance. The AI tool may oversimplify by eliminating advanced features, assuming they are “too complex” for someone with a disability rather than allowing the person to choose. 

​Cost can be a factor as many AI-powered tools carry a hefty price tag. We also know that a digital divide exists. Some people may have limited or no access to the internet, newer technology, or tech training which may make using AI difficult or impossible. 

​Another major concern is privacy and security. AI tools are usually cloud based and we must limit sharing private, personal data. Insecure systems may leave the user vulnerable to data breaches, misuse of information or exposure to unwanted marketing. 

​Considerations when incorporating AI with AT 

​As we think about the future of AI as a tool for people with disabilities it is important that inclusive design and ethical considerations are at the forefront. Even more importantly, there must be active involvement of the disability community in the development of the process.