Tech accessibility: Inclusive design, ableism, and ethical AI

To follow inclusive design principles, Dona recommends giving everyone the same opportunities at work, such as attending an event or turning on captions, and let them decide whether you accept it or not. “Don't be biased based on some internal notion that people with disabilities don't want to attend an event, represent your company at an event, or that they may not even be able to get there on their own, etc. Let them figure it out. Don't infantilize people with disabilities with their own ableist thoughts,” she explains. 

Dona also notes that people with disabilities shouldn’t have to ask for accommodations in every single meeting. Those accommodations should be assumed with disability-first work as the standard.

When it comes to meeting accessibility standards, Dona stresses that “you can’t just talk to one type of disabled person” and that it should start from the top. She uses the example of building a new product. No matter whether it’s a gas station or an AI super drone, people with disabilities should be included from the beginning to make sure basic requirements are understood. 

Next, as you’re working toward inclusive product design and user experience, figure out how someone who uses a screen reader, someone with limited mobility, or a colorblind or dyslexic person would use this product. As Dona says, don’t make stuff up if you want to design for accessibility. 

Then, make sure you have a set of trusted advisers that you lean into across the disability spectrum. Product managers and developers should add accessibility to the product development process.“For example, as you’re doing a design review of your code, make sure someone is raising their hand and asking about accessibility,” says Dona. “Where are you adding labels, names, ARIA factors, alt text…? Put it as part of your pull request and CI/CD process as an automated check-in that asks if you tested for accessibility.” 

Just as you consider security from the beginning of the product development process, you should also consider accessibility from the start. That way, it’s not a huge added task; the processes are already in place with product development. 

There are three key things Dona recommends organizations keep in mind when building inclusive workplaces. First, make things available to everyone, not just for the people who ask. 

Dona gives the example of captioning. Instead of using a virtual video conference software that requires attendees to ask to turn the captions on in front of everyone, organizers of these meetings should turn on captions by default for greater accessibility. Attendees who don’t want captions can choose to turn them off. 

Next, consider diversity in the workplace and how it relates to the consumption of content. “If I’m presenting content at all, I send out those slides or a link to the content before the meeting. That way folks who might be introverts, have a learning disability, or may be blind can consume that content earlier and know what will be presented in case it’s not presented according to accessibility standards,” explains Dona. 

Finally, be sure to record any meetings and send out the recording and the transcript aftward. That way, anyone can catch up on things they may have missed in the meeting. 

One of the biggest missteps Dona sees leaders make when it comes to accessibility and inclusion is assuming what people can and can’t do. Business leaders may assume someone may not want to attend a conference because they’re in a wheelchair, are blind, or are a new person. 

This assumption limits opportunities for individuals and fosters greater externalized ableism. Dona shares how to combat this assumption: “I've started to tell leaders to give the same opportunity to everybody at your organization who is qualified and let them decide to opt in or opt out. That is just a basic thing that leaders can do.”

It’s not uncommon for those in accessibility communities to experience imposter syndrome due to internalized and external ableism. Even those who advocate for diversity and inclusion in the workplace can feel like they’re not qualified to do a job or lead a particular project. 

Dona shares her own experience with imposter syndrome while presenting: “I felt this overwhelming need to memorize everything and not mess up anything. I realized it was my own internalized ableism that I had to come to terms with. I assumed I had to present like a non-dyslexic person to be heard, and I had to confront that internalized ableism first. Then, I had to confront the fact that it is absolutely okay not to know how to do something the first time.”

External ableism also impacts feelings of imposter syndrome as it only encourages insecurity around abilities. “I’ve experienced people thinking I can’t do something which then feeds the internalized ableism. Then, you’re dealing with the external assumption about what you can do and the internal assumption that you don’t know if you can do that,” explains Dona. “People are telling you that you can’t; those assumptions don’t help us get over impostor syndrome.”

Externalized ableism can lead to organizations sticking individuals in what Dona refers to as the “accessibility box.” Accessibility isn’t just a matter of hiring disabled people for accessibility programs. People with disabilities are capable of working in departments across an organization. In other words, it’s important to include people with disabilities in core functions, such as product management. 

Dona explains, “I want people to work in product, but not just accessibility. I love when I meet a blind person or someone in a wheelchair or with dyslexia, and they tell me I write the kernel for Windows. By default, this product will be more accessible.”

Dona believes that AI can be an amazing tool for people with disabilities to level the playing field in life and at work when used correctly. However, there are three considerations to keep in mind when it comes to AI ethics and accessibility. 

First, there’s the basic accessibility of AI products. AI products must be accessible from the start. For example, a screen reader user should be able to type words into a box or field and read what is being said. 

Second, is the AI product actually usable for people with disabilities? Continuing with the example of a screen reader, the paradigms of generative AI are quite different from other things we have used. Dona asks, “Does the screen reader user understand where the words are being generated from and how accurate they are? Are there sources? Is it clear that these are the sources the content is referring to?”

Finally, AI ethics come into play. The datasets used for AI are biased against people with disabilities, in emerging markets, older people, people from different cultures or countries, etc. AI is largely skewed toward a certain demographic. “That’s the bias of the world we live in. We need to have more disability representation to work in AI products, whether it's training models, working with the moderation, or weighing certain datasets higher to represent people with disability better.”

For a more ethical AI, people with disabilities must be involved in the creation process from all sides. Furthermore, responsible AI practices are paramount. Always state that content is AI generated when applicable. Dona recommends making sure your product is doing exactly what it’s supposed to be doing and nothing else to meet AI ethics. Likewise, cite the sources content and activities are generated from. 

Finally, don’t even consider using ChatGPT to replace people with disabilities, which Dona describes as the absolute worst application for the generative AI tool.

You catch Elsa and Dona’s engaging conversation and more insights on how accessibility can help create better products for everyone in the full Expert Access episode on YouTube.