Putting AI to work in accessible tech

Born profoundly deaf, I’ve always worn hearing aids. And over the years, I’ve seen this technology totally transform. What started as old-school ear trumpets evolved into body-worn aids, behind-the-ear and in-ear tech, and the digital age of rechargeable batteries, cochlear implants, and wireless devices with AI features. They’ve gone from being huge, clunky things to super-tiny smart gadgets. How I felt about wearing them has also changed. Once I felt awkward. Now I feel much more confident. Through both small steps and big leaps, the tech has steadily improved the lives of people with hearing loss.

Both in the world of work and at home, I live and breathe captions and have watched this tech evolve too. From the early days of basic, delayed text to today's AI-powered, near real-time accuracy, the advancements have been remarkable. The recent AI-powered ability to translate captions into 140 languages also provides a global reach. Captions enable me to participate in meetings fully and confidently engage in phone calls with colleagues and clients. My experience of various forms of assistive tech, throughout my education and in my work life, has shaped my understanding of inclusivity, which I now help implement in the workplace.

But where will things go next? I’m really excited by the possibilities AI presents for the future. And as a Thoughtworker, I’m also aware that technology doesn’t spring forth in perfect form from a vacuum. The intersection of AI and accessibility will hopefully bring incredible advancements, but only if managed thoughtfully and carefully.   

AI is already proving its worth

My own experiences aside, AI-based tech is making a major difference to people’s lives. Welcome to a world of text-to-speech (TTS) features and AI-powered smart glasses. There are apps like Seeing AI, which uses machine learning to recognize objects and describe them to a user through the phone’s camera in real-time. While ‘Be My Eyes’ connects people who need sighted support with volunteers and companies anywhere in the world, through live video and artificial intelligence.

Prosthetics are also improving. Artificial limbs can now ‘learn’ individual movement patterns and adapt to provide smoother, more natural control, improving mobility and functionality. Meanwhile, AI-powered wheelchairs navigate obstacles autonomously and respond to user voice commands, offering greater independence and freedom. AI can also assist with cognition, with AI-powered assistants to help manage daily tasks and schedules, providing valuable support for people with memory impairments or executive function challenges.

AI as an accelerator

As well as being a key feature in many accessible products and tech, AI also has the potential to significantly accelerate accessible tech design and development. It can be used to automatically test applications and websites for accessibility issues, streamlining the process and making it more efficient. It can also be used to analyse user data and preferences to recommend accessibility features and personalize designs for diverse needs.

Software development relies on a solid understanding of user needs, and AI can help here too. We can put it to work to analyse user interactions and feedback to identify unmet needs and inform future design decisions. And when it comes to writing code, AI works at impressive speeds. Give an AI-coding assistant a set of accessibility guidelines, and it can generate code automatically, ensuring it adheres to your standards from the outset.

All this speed and efficiency are worthy of praise, but AI does have its limitations, and ultimately it’s just another tool in the developer’s kit. Its impact depends on how we use it.

Understand AI’s limits. And take the right precautions.

It's a myth that you can rely on AI for all your accessibility needs. It simply can’t ensure total compliance on its own. It sometimes gives wrong answers and commands. And despite its potential to remove human bias, algorithms are still susceptible to the biases found in their training data and can be discriminatory.

But being aware of these issues is the first step towards mitigating them. The next step is putting thorough processes in place to avoid potential pitfalls. Things like ensuring inclusive data collection and algorithm design to avoid bias. Having a strong ethical framework to guide the development and development of AI for good. And building transparency and explainability into AI systems to ensure accountability and trust.

Collaborating with people with disabilities in the design, development and testing phases is vital to create truly effective and beneficial AI-powered accessibility solutions. And, as a foundation to all of this, take a human-centered approach: AI should be used to support and empower human designers and developers, not replace them entirely. Remember, AI is not infallible, and humans are prone to accept its authoritative responses.

Spreading the word

With principles like these in place, developers can maximize the advantages afforded by AI and bring advances in accessible tech to market sooner. Raising awareness around inclusion is a key part of my role at Thoughtworks. By bringing the value of inclusive development up the agenda internally, and for our clients, I’m helping to facilitate the creation of new technologies that will really make a difference to people’s lives, as they have done to mine.