Accessibility has been the catalyst for some of our greatest leaps forward. Take, for instance, the introduction of graphical user interfaces (GUIs) in the 1970s, which made computers more intuitive, user-friendly, and accessible to a wider audience. Accessibility sparks innovation because it redefines usability. Combined with AI, it can help solve even the most persistent problems.

Here are three stubborn problems AI and accessible design can help solve.

Problem No. 1

The sea of sameness.

Most new products are just more of the same—and for good reason. Originality is expensive, requiring time-consuming prototypes for testing. Too many variables risk the bottom line, trapping most new product designs in an echo chamber. The same problem creates a shortage of assistive technologies.

A radical shift occurs when we design for disability first; we gain new ways to deliver disruptive solutions by designing for the margins.

When we design with the same parameters and methods, the results are predictable (great) and similar (not so great); this way of working encourages sameness. A radical shift occurs when we design for disability first; we gain new ways to deliver disruptive solutions by designing for the margins.

Digital distractions and natural aging reduce acuity, carrying a load or suffering an injury limits mobility, crowded, noisy rooms or too many concerts cause hearing loss. We’ll all experience disability in one form or another, whether permanent or temporary, and need flexible products and services that cater to our needs and are easier to use. Intentionally designing for sensory, mobility, or cognitive differences serves everyone.

So, how do we create breakthrough products for everyone without breaking the budget?

Consider how 3D printing slashed the time and cost of physical prototypes. Small parts could suddenly be made in the office and on demand rather than hand-machined and shipped. Similarly, generative AI allows designers to quickly create visual prototypes for stakeholders, investors, and focus groups. Lower development costs mean businesses can afford to explore edge-case scenarios informed by accessibility insights. Combining AI and accessible design gets more creativity into the product pipeline; this is important because more creativity in the pipeline is a pathway to product differentiation.

Problem No. 2

Privacy or personalization? Pick one.

Today, every system requires a unique login and profile (unless you're willing to give up your data to a company via single sign-on), leaving users stuck managing multiple logins for different accounts. People with disabilities have the added chore of scouring settings to adjust each platform to meet their needs. Luckily for everyone, the password era is on its way out. Biometric continuous authentication tools like Cisco Duo, which protect intellectual property and regulated data, posit a universal digital login with one-way encryption security while retaining personal data ownership.

Moreover, personalized generative user interfaces, such as Vercel AI SDK 3.0, can tailor systems to specific individuals. So, no more manually turning on subtitles or closed captioning for every video. Forget about forms to request help with a wheelchair or assisting a parent with young children. AI can apply our specifications en masse to system capabilities for turnkey personalization, while a public universal login protects privacy and eliminates the need for countless distinct passwords.

Problem No. 3

Voice UI is stuck in beta.

For something to be considered usable, it has to meet three minimum requirements. It must be visible, error-free, and efficient. A screen and keyboard setup meets this criteria. Voice UI does not. So far, replacing a keyboard or screen with voice is confusing, frustrating, and slow by comparison. The experience of home smart speakers, phone voice assistants, and screen readers for blind people is second-rate—a beta version of voice UI’s promise at best.

We can overcome the limits of voice interaction usability using AI tools, such as small language models.

The problem is that voice systems today are built with discreet libraries of boolean commands; they aren’t set up for complex queries like conversation. Speech is based on context, not logic—the foundation of boolean commands. We can overcome the limits of voice interaction usability using AI tools, such as small language models (SLMs). Even with the 2.7 billion parameters of Microsoft’s Phi-2, an SLM’s size is more conducive to working with natural language. An SLM can make interactions faster by skipping ahead and automating multi-step tasks on command while also monitoring task speed and errors for auto-correction.

Voice UI holds incredible potential for users with limited vision or mobility and hands-free products for all. With an SLM, voice could be useable enough to be trustworthy at scale.

So, what’s next?

These examples challenge the myth of accessibility as a loss leader. The bottom line is that accessibility is good for business. By embracing accessible design and AI, we can drive innovation in unexpected ways, reach the next level of human-centered design, and create products and systems that are not only more feasible, usable, and desirable but also serve everyone.