How AI is changing accessibility

At this point, ‘paradigm-shifting technology’ has been used to describe AI so frequently that the words are losing meaning. But it’s worth taking a moment to reflect on how true it is. New and emerging technologies are upending previous assumptions and approaches to how we live our daily lives—for better and for worse.

One of these ongoing shifts, and something that may continue to change drastically, is accessibility. At its core, ‘accessibility’ is the practice of making information, activities, events, or environments sensible, meaningful, and usable for as many people as possible. By automating previously time-consuming processes, analyzing large quantities of data, and continuously adapting as they recognize patterns and make decisions, AI is upending previous assumptions about what it means to make something accessible.

What is accessibility?

Thinking about accessibility begins with considering what might make something inaccessible. Someone with no or low vision might have a harder time with a website if the images have not been captioned with alt text or image descriptions. Someone working in a loud café might have a hard time understanding a video without closed captioning.

Fundamentally, disability describes a mismatch between a person and their environment. Someone working in loud location would be experiencing a situational disability, while someone with low vision might be experiencing a permanent or temporary disability, but they are both experiencing a mismatch in the environment that interferes with their ability to perform a task.

Regardless of the type of disability, it’s important to consider what might make something inaccessible, so that you can begin to think of ways that it could be changed to become accessible.

Consider the following four principles of accessibility from the Web Content Accessibility Guidelines:

• Perceivable - Information and user interface components must be presentable to users in ways they can perceive (it can’t be invisible to all their senses).
• Operable - User interface components and navigation must be operable (the interface cannot require interaction that a user cannot perform).
• Understandable - Information and the operation of user interface must be understandable (the content or operation cannot be beyond their understanding).
• Robust - Content must be robust enough that it can be interpreted reliably by a wide variety of user agents, including assistive technology (as technologies and user agents evolve, the content should remain accessible).

Why is accessibility important for customer experience?

A curb cut is a lowered section of sidewalk. Originally designed to make it easier for people in wheelchairs to cross the street, curb cuts are frequently used by cyclists, skaters, rollerbladers, and pedestrians pushing strollers or carts or walking with small children. The ‘curb-cut’ effect describes the phenomenon of disability-friendly features being used and appreciated by a much larger group, outside of the group they were originally designed for. Often, these curb cuts are widely adopted as a general convenience that benefits everyone, rather than being viewed as an additional accessibility device reserved for some.

The electronic curb-cut effect

TV manufacturers have taken note that the caption decoders that enable closed captioning for hard-of-hearing viewers wound up benefiting millions more consumers, simply because many more consumers choose to use closed captioning while watching tv, regardless of hearing ability. Closed captioning makes it possible to watch and understand a show in a noisy bar or in a silent library. Viewers learning a new language can see new words written out as they are spoken on screen, or have captions in their native language to aid understanding.

Generally, televisions with decoders that enable closed captioning work better for everyone, regardless of hearing ability. And accessible customer experience is a better experience for everyone.

Like a rising tide that raises all ships, designing with accessibility and disability in mind improves the experience for everyone.

How AI is improving accessibility today

AI is an incredibly powerful tool for providing personalized services and meeting individualized user needs at an unprecedented scale. But in order to capitalize on these benefits, organizations will have to consider accessibility at every stage, by including diverse user research and considering inclusive, human-centric design.

1. Many technological advances and innovations have been developed by, or in support of, people with disabilities. In light of this, the Start with One, Invent for Many experiment from Google pairs an engineer or designer with one person to create an innovative solution using Google products. The hope for this ongoing project is that a solution intended to help one specific person can end up benefiting many, many more people.
2. Signapse uses synthetic signers, or AI-generated sign language interpreters, to provide sign language translation in places where it might otherwise not be available, such as transportation, and is currently being used at Huddersfield Railway Station to provide real-time information on updates and cancellations.
3. Be My Eyes, the mobile app that allows anyone to assist visually impaired people through live video calls, recently introduced the beta release of Be My AI. Powered by OpenAI’s GPT-4 language model, Be My AI offers AI-powered visual assistance for instantaneous image-to-text generation. Users can send images via the app for instantaneous identification, interpretation and conversational visual assistance for a wide variety of tasks.
4. Government services in Nunavit, Canada are now using an AI solution to provide information in in Inukut, the indigenous language, opening government resources to a larger group of people and helping to revitalize the indigenous language.
5. Some US cities are using AI to make emergency services more accessible. Presidio County in Texas previously spent precious minutes calling a third-party interpreter if a dispatcher did not speak the same language as a caller. A new AI-enabled translation service provides real-time transcription for multiple languages, lowering the barrier to emergency services for the 77% of residents who speak a language other than English at home.
6. Another Google product, the android app Look to Speak, allows people to use their eyes to select pre-written phrases and have them spoken aloud. The app is now available in 19 languages and was recently released in Ukraine to help refugees facing even more barriers to communication.
7. Deliveroo utilized TTEC Digital and AWS to overcome language barriers and make their food delivery platform more accessible. By designing a contact center platform that supports live chat with built-in automatic two-way translation, Deliveroo was able to increase their first call resolution to 83%.
8. Voice-enabled interfaces make it possible for more people with a broad range of disabilities to interact with generative AI. Google’s Contact Center AI (CCAI), with Dialogflow, a natural language understanding platform, has built-in integration with Google Cloud Speech-to-Text API, making it possible for developers to create chatbots that support voice-enabled input. CCAI can analyze multiple types of input, including text or audio, and respond through text or synthetic speech.
9. The use of Google’s Image description technology reduces the efforts of software developer's insertions of descriptive text for images.

When AI interferes with accessibility

While AI can be very useful for equalizing access or removing human bias from the equation, it can also have the opposite effect. As we’ve written before, AI is extremely useful for recognizing patterns in massive pools of data and generating content or drawing conclusions, but can include patterns we find less desirable, like human biases, prejudices, and mistakes. To ensure we don’t further entrench existing disparities, AI can’t operate in a black box.

The U.S. Equal Employment Opportunity Commission (EEOC) is working on ensuring that the use of AI software in employment decisions complies with federal civil rights laws — especially important now, considering that nearly 76% of organizations use algorithms to assess performance on hiring tests and 40% use artificial intelligence (AI) when screening potential candidates, according to Disability:IN. As Chair of the EEOC Charlotte Burrows phrased it, “We totally recognize that there's enormous potential to streamline things, but we cannot let these tools become a high-tech path to discrimination."

Including accessibility as part of the CX design

Companies that ignore accessibility principles do so at their own peril. According to the Centre for Inclusive Design, “the relative cost of retrofitting a product or service to become inclusive will increase significantly over time and can reach up to 10,000 times the cost of making it inclusive by design.” With the significant buzz around new AI capabilities, and the corresponding pressure to release more new products faster than the competition, it’s a potent reminder that moving thoughtfully can pay more dividends than simply crossing the finish line first.

Previous accessibility advancements like curb cuts and closed captioning offer concrete examples of how designing accessibly is not only the right thing to do, it’s also the profitable choice. Improving accessibility improves the experience for everyone — consumers and companies alike.