How AI Is Improving Web Accessibility

With each passing year, Artificial Intelligence (AI) is growing at an ever-increasing pace. What was once considered possible only in science fiction is now making waves in computer science and the workplace. From smart appliances to smart cars, it continues to make new strides in almost every aspect of our lives, especially in the digital world. 

However, one of the lesser-known benefits of AI is web accessibility. Various AI-based technologies are making the digital world more accessible to people with disabilities. AI is helping disabled people interact with the Web just the way everyone else does.  

AI-based solutions are proving to be a game-changer in the time when the risk of an accessibility-related lawsuit is on the rise. In 2018, the number of federally-filed website accessibility cases skyrocketed to 2,285, up from 815 in the year prior. In the first half of 2019, total filings reached over 3,200, breaking all previous records. 

Let’s take a look at some of the ways AI is already changing the world of website accessibility. 

Language Translation and Captioning

The use of Automated Speech Recognition (ASR) algorithms based on AI technology is creating new possibilities to create captions and subtitles for online video content. These technologies are helping users who are deaf and hard-of-hearing to interact with video content. 

Not only can these computer programs provide live captioning, but they can also improve their performance by self-learning. So, an ASR system can learn to deliver better captioning over time. 

Several leading technology companies are also working on developing various language translations and captioning solutions for the disabled. As part of its goal to create a more inclusive Web, Microsoft has built Microsoft Translator, an AI-powered communication technology, for the hearing-impaired. 

Available in more than 60 languages, Microsoft has partnered with several educational institutes as well to improve access for disabled students. About 1,500 students who are deaf and hard-of-hearing that are an integral part of campus life at the Rochester Institute of Technology are taking advantage of this technology.

Another exemplary AI-based technology that is helping improve web accessibility is Google Translate. With the new Google Neural Machine Translation (GNMT) system, Google has managed to increase the accuracy of its real-time translator. 

GNMT reduces translation errors by more than 55%-85%. The system also gave rise to the concept of sentence-by-sentence and idea-by-idea translation instead of word-by-word translation. Although the technology is still not perfect, it does represent a significant milestone. 

Automatic Image Recognition

Whether it is Instagram or Facebook or even blogs, images are an integral part of the Web. Unfortunately, visually-impaired and blind users are unable to view and understand the reference of these images. However, AI-based image processing is helping visually-impaired users recognize the images. 

Google has developed an advanced algorithm called Google Vision API, which uses neural networks to recognize images. With this tool, blind users will know if the image is of a cat or a human. It also has the ability to determine if the image can appear in the safe search category. 

Facebook had taken similar steps when it launched its image recognition API in 2016. Powered by neural networks and machine learning, this technology can describe images for blind and visually-impaired users. 

However, you will still need to use Alt Text because this technology is still in its infancy. It will take a few more years until APIs can accurately describe images. 

Facial Recognition

In most cases, you have to use a password or a four-digit pin to log in to your laptop or phone to access the internet. With AI-based facial recognition software, however, web accessibility comes easily to blind and visually-impaired users as it eliminates the need to use passwords. 

Many website developers and tech companies are also experimenting with facial recognition as an alternative to online CAPTCHA. Visually-impaired and blind users often find it difficult to fill out the CAPTCHA code, which is an essential security measure on many websites to avoid bots. 

Information Summarization

Although the Web is full of images and videos, it still largely comprises text content. While it is easier for visually-impaired users to interact with text content using screen readers, reading lengthy documents, articles, and blog posts are not easy.

Many companies are using AI-based algorithms to create shorter text summarization of voluminous articles. It can help break down lengthy and complicated information into digestible chunks for blind and visually-impaired users. 

Salesforce has been at the forefront of developing this technology. The company has made new breakthroughs in natural language processing. The software uses a combination of a more contextual word generation model and Reinforcement Learning (RL) to provide highly accurate and readable text summarization. 

Automatic Lip Reading

Similar to creating live captioning for video content, an automatic lip-reading algorithm helps you provide web access to people with hearing disabilities. Google’s DeepMind is one of the best examples of this technology. 

DeepMind, which was developed based on more than 5,000 hours of TV including shows such as Newsnight, BBC Breakfast, and Question Time is capable of identifying which lip movements relate to which words. In the pilot test, out of the 200 randomly selected clips from the data set to decipher, the AI correctly translated 46.8%.  

The algorithm can help deaf and hearing-disabled users with real-time speech into text translation with higher accuracy. However, this technology is still being perfected to produce better results. 

Another major stride in the voice-recognition or speech-recognition field is Google’s Project Euphonia. This AI-based technology is helping computers to decipher various human speech patterns to create better transcriptions for the disabled. It can help people with impaired speech or who cannot speak to access the Web with hand gestures and sounds instead of perfect speech. 

The Future Looks Exciting 

AI is one of the fastest evolving technologies on the planet right now. The inventions mentioned in this article are just the beginning. In the coming years, we are likely to see many more inventions that increase the performance of assistive technologies.

As you already know, most websites and online shops have been at the receiving end of the ADA lawsuits. AI-based assistive technologies present an easily scalable and affordable way to overcome such issues. 

Furthermore, these technologies can help you spread the word about the need for web accessibility and how easily it can be incorporated. Soon, we will have tools that can turn any website, even a huge e-commerce site, into a completely accessible one in just a few clicks.

Parting Words

As you can see, various AI-based technologies are being introduced that are bound to make the Web more accessible for the disabled. While many of these inventions are still in their infancy, developers, technology experts, and even disabled users can see their potential to make the Web more inclusive than ever. Let’s hope, in the coming years, AI may even help make the Web completely accessible. So, you need to start incorporating accessibility into your website to stay ahead of the curve.

What do you think?