Members of the disabled community cheered earlier this year, when voice-enabled AT&T-calling with Alexa arrived. This new functionality solves an immediate need for anyone who has to reach a phone quickly during an emergency…or simply answer a call from a loved one.
It’s just one example of the power of artificial intelligence (AI) and machine learning (ML) to improve life for those living with disabilities, and its utility will no doubt expand as people use it to solve previously insurmountable challenges.
Voice-enabled calling joins a host of new and emerging technologies that are poised to transform accessibility with the help of AI and ML. Just imagine the life-altering impact of:
- Highly accurate auto-captioning for those who are deaf or have hearing loss
- Facial or image recognition for people who are blind or those with low vision
- Natural language understanding to facilitate comprehension for those with cognitive disabilities
- Autonomous cars for individuals otherwise unable to drive
However, to be truly accessible, AI-enabled solutions, like any technology, must be consciously designed for use by people with disabilities ― from the ground up. In the case of AI and ML, which rely on data sets to train "intelligent" models, data scientists can play a crucial role. By choosing to use inclusive data, they avoid a potential source of bias.
Dr. Aaron Bangor, a co-founder of AT&T’s first company-wide group dedicated to promoting accessible technology, urges organizations to get in front of the data bias issue, particularly when it comes to powerful technologies like AI. To do so, he recommends adopting a comprehensive approach throughout the research, design, and development process. “If we want to create equitable outcomes that serve the full community, we have to call out the risk of inherent bias,” he says. “It’s imperative.”
To mitigate the effects of bias, Aaron says technology development should be inclusive from end-to-end, accounting for how data are collected and how algorithms are trained, as well as the design and development of AI-enabled solutions. This concept takes an organization’s commitment to accessibility to new levels, an approach he calls “Enlightened AI.”
He points to three factors upon which Enlightened AI depends.
1. Human-centered, universal design. “We must recognize that universal design benefits everyone – and view it as a catalyst for inclusive innovation, not simply another hoop to jump through,” he says. “This includes a focus on what the technology does and how it meets people where they are. The AT&T Accessibility Awareness Lab brings to life the idea that we have to get out of the mindset of designing for ‘people like me’ – even subconsciously – to think about a wider user base that enables everyone.”
2. Diversity of perspectives at the table to fuel innovation. “Design is most inclusive when people with varied experiences, including the impact of accessibility barriers, are directly involved from ideation through development. We must work with individuals who have traditionally been left out or discounted by standard design practices. For example, a blind web developer automatically defeats the notion that web sites are just for people who can see a screen. Similarly, by including data scientists with disabilities in data collection, algorithm development and application development, we can prompt greater awareness of potential pitfalls and opportunities for positive outcomes.”
3. Investment in diverse education at all levels. “Ultimately, humanity and empathy come from people. We need to make sure those who represent diverse abilities and perspectives are ready and equipped to contribute meaningfully. That takes a long-term plan to get more people with disabilities into technology, AI and ML ― from kindergarten through grad school,” Aaron says. “As long as education has accessibility barriers, people with disabilities will be systematically filtered out of technical fields. For business success and a robust society, it’s critical that we build a pipeline to prepare future, diverse talent.”
“It’s not about doing it for ‘others,’ it’s about doing it for all of us. That’s enlightened AI.”