Recent advances in generative AI could revolutionize assistive technology. For people relying on assistive tools, AI-powered devices could usher in a new era of accessibility.
More than 2.5 billion people require assistive technology to help them navigate the world.
The World Health Organization predicts that the number of individuals with cognitive, sensory or mobility impairments who need assistive technology will rise to 3.5 billion by 2050, due in large part to a growing number of older individuals. AI -- in particular, generative AI -- is expected to help meet that growing need by making those assistive devices more useful and powerful.
"Accessibility will be truly revolutionized by AI," said Eshed Ohn-Bar, an assistant professor of electrical and computer engineering at Boston University and a Hariri Institute junior faculty fellow.
AI, which is already enhancing existing assistive devices, is also expected to create new opportunities for users. AI-enabled devices could assist visually impaired individuals, serve as communication aids for people who have difficulty speaking and help those with dementia complete daily living tasks.
7 examples of AI-enabled assistive technology
Revolutionary advances using AI are already happening.
For instance, AI recently helped a man with paralysis walk again. In May 2023, AI technologies embedded in a 40-year-old Dutch man's body to transmit thoughts along his nervous system enabled him to walk after being paralyzed in a biking accident 12 years prior.
Ohn-Bar and other experts in the use of AI with assistive devices said the majority of opportunities that AI will bring are only in the early stages. Even so, there are multiple examples of how AI can transform assistive devices and the lives of individuals who rely on them.
1. Detection and response
Devices using AI can monitor individuals to determine in real time whether they need help and what help they need, explained Karen Panetta, an IEEE fellow, professor of electrical and computer engineering at Tufts University, and dean of graduate education at Tufts' School of Engineering.
As an example, she cited a device that IEEE members helped develop after data analysis determined that autistic students' attention to writing dropped off when the pressure they applied to their pens waned. In response to that finding, researchers developed a prototype that detects when the pressure drops and then alerts students to that change so that they can refocus.
2. Advanced environmental guidance
AI can help individuals with sensory disabilities achieve greater independence by guiding and assisting them in nearly any environment, Ohn-Bar said.
Ohn-Bar's work focuses on developing AI technology that can provide humanlike assistance to those with visual impairments. For example, using AI alongside other technologies, such as sensors, can provide visually impaired people with audio instructions that help them more safely navigate their surroundings.
The AI system can adjust guidance in real time based on various factors to ensure it's delivering the right type and amount of information for each individual, Ohn-Bar said. For instance, using machine vision to measure depth enables the AI system to alert a visually impaired person to the presence of stairs or an uneven surface.
3. Assistance with daily life
Organizations are creating devices, including robots, that can assist individuals with disabilities with their daily needs. For example, researchers are building robot companions that can provide medication reminders, pick up items or offer companionship to people in cognitive decline.
"Many people think of these robots as toys, but they're interactive and they're used as companions," Panetta said. She noted that while pilots of such technology have produced mixed results in terms of user acceptance and accuracy of actions, she is confident that technology advances will create better robots more capable of successfully meeting user expectations and needs.
AI is also helping to create smart homes that serve people who use assistive technology, enabling aids such as voice assistants and responsive systems that adjust automatically to the occupant's requirements for safe and comfortable living.
4. Mobility aids
AI is transforming how devices help those with mobility issues, Panetta said.
For example, AI could help improve physical therapy. An AI device could analyze an individual's movement, such as how they distribute weight while walking, and then give feedback on how to adjust their movement to improve physical ability.
AI-enabled hearing aids are one example. Hearing aids can use AI to distinguish between sound types and then modify the sounds to improve the user's experience, Panetta said. AI enables these devices to suppress distracting background noises and enhance conversational sounds, improving the clarity of sound delivered to the user.
The AI-enabled device does this dynamically, Panetta said, adjusting as the user moves from one auditory environment to another. Over time, the AI model learns from these experiences to continually refine its sound delivery.
Moreover, AI can bring more nuanced, accurate and responsive capabilities when applied to technologies such as sign language recognition and text-to-speech devices.
6. Robotic movement
Although AI has been used in prosthetics and mobility aids for many years, advances in machine learning algorithms for robotics have enabled more accurate analysis of data from sensors in prosthetics and mobility aids. AI models can also respond and adapt to input in order to guide users or create more natural movements.
AI can help technology meet users where they are.
Mike BechtelManaging director and chief futurist, Deloitte Consulting
7. Advanced experiences
Mike Bechtel, managing director and chief futurist at Deloitte Consulting, sees a future where AI can create experiences that diminish the barriers that some disabilities create. He cited the potential use of technology to detect brain activities that AI can turn into actions, which could enable individuals to control their actions through gestures or even thoughts.
Such capabilities are far off in the future, Bechtel acknowledged. But in the near term, AI could create advanced artificial and virtual reality interfaces that strengthen users' connections with the world.
"We're able to cast a dramatically wider net to serve users who may have been boxed out by an interface that requires sight or hearing or other [physical or cognitive] capabilities that the user might not have," Bechtel said. "AI can help technology meet users where they are."
The benefits of AI-enabled assistive technology
There are numerous examples of AI-enabled devices and systems that could change the landscape of assistive technology. Integrating AI capabilities could offer benefits for a wide range of assistive devices.
Personalized assistance
AI models, with their ability to learn from historical data patterns, can personalize systems and experiences to each user's needs. This enables assistive systems to fine-tune themselves to each individual rather than deliver a general, one-size-fits-all option to everyone.
"There are different levels of ability, and one size doesn't fit all," Panetta said. "The current thinking is to use AI to tailor [technology] to best meet a specific user's needs."
Continuous improvement
Similarly, AI systems' ability to adjust their behavior in response to changing data also helps them improve on an ongoing basis, Ohn-Bar said.
For instance, an assistive technology can learn the unique patterns of a user with a speech impairment so that it produces increasingly accurate results over time.
"This personalization will be automatic with ChatGPT-type technology," Ohn-Bar said.
More readily available assistive tech
Generative AI coding tools are enabling software developers to create applications and features more efficiently.
"What we're seeing with generative AI is that the amount of effort that sits between idea and implementation is greatly shortened," Bechtel said. Developers can use AI, particularly generative AI, to create code at a speed and scale that humans alone can't match, he explained.
This speed and scale could translate to more efficient development of assistive features. If accessibility-related features are easier and cheaper to build into applications and software, developers might be more likely to integrate them into new and existing systems, thus making assistive technology more readily available to those who need it most.
User-created assistive tech
Nontechnical users could even use generative AI to develop their own assistive functions and features. Because generative AI and other no-code tools make development more accessible, Bechtel said, individuals can create their own tools or modify existing ones.
Large language models such as ChatGPT could also help users to tweak a system in ways that work particularly well just for them using natural language requests. "Users can tell the system what exactly they want and make a system that works for them as individuals," Ohn-Bar said.
Mary K. Pratt is an award-winning freelance journalist with a focus on covering enterprise IT and cybersecurity management.
