Internet accessibility can be a major problem for the visually challenged. If a website doesn’t support screen readers, visitors with sight problems have to get help from someone else to use the site.
Given that most businesses or organizations, from healthcare to hospitality, primarily use websites to interact with their customers, you can understand the issue. Though the Americans With Disabilities Act (ADA) allows individuals to sue businesses for inaccessibility, the US Supreme Court appears poised to overturn that option, which could create a variety of issues related to information access.
Since generative AI is based on large language models (LLMs), it can read websites and interact with users verbally, making it the potential fix for information access. (Physical access is a problem that might be addressed by autonomous wheelchairs or improved electronic prosthetics, but that’s a topic for another day.)
Information access and generative AI
The internet is a visual medium. Even audio files are found and accessed visually. The first generation of digital assistants such as Alexa and Siri are verbal browsers that convert questions into web queries and then read back the results. But those tools are inadequate for anything outside of simple responses to questions. They cannot fill out forms or even read or navigate web sites.
Generative AI, which consumes information on websites and can respond to far more complex commands, is a big step beyond those basic assistants. Additionally, as generative AI advances and learns about its users, it can autonomously perform roles that would otherwise require a human touch, which would be convenient for sighted workers, but critical for workers with vision loss.
This could lead to an interesting outcome.
Gen AI could level the playing field
The need to use a tool such as generative AI will be far greater for those with vision problems, so once they’re provided with it, these workers should become proficient quickly. This proficiency, particularly for those working remotely (who do not have to navigate to and from work), should not only massively improve their own productivity but could allow them to exceed sighted peers who aren’t using AI routinely — or at all. Note: generative AI is already being used to enhance accessibility and inclusion for people with disabilities.
This suggests that as AI tools advance and people use it more and more, there’s a good chance that, at least initially, the combination of AI and visually impaired workers will be greater than the sum of the parts.
In short, the people that will most benefit from generative AI should become examples of what blending people and AI can do; it might well result in tools that partially block sighted employees’ vision for them to better focus on the AI interface and the task at hand. This scenario could be one of the first examples of how AI can enhance people (and productivity) by tying them closer to technology and forcing a more integrated bond between employees and the advanced tools they use.
It does look as if the Supreme Court is about to cripple many ADA rules that the visually impaired feel are critical to their ability to function in a sighted world. But generative AI could be the tool that at last addresses their work needs and provides them with a significant incentive to become AI experts. That expertise and enhanced productivity won’t just allow them to match their colleagues’ performance but could help them advance their careers though the motivated use of these tools. We’ve already seen that these tools can increase productivity by between 30% and 80%. For the visually impaired, that performance jump should be far more significant.
In the end, one of the big benefits to AI in general may be leveling the playing field for those of us that have impairments — and generative AI may have the biggest impact on the visually impaired than on any other group.