BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

How AI Is Being Used To Help Blind Students ‘Visualize’ Graphs And Charts

Following

Roll back the decades, or for the more youthful, a shorter few years, and reflect for a moment on the foundational educational importance of being able to interpret graphs and charts from early high school years onwards.

Imagine then, the experience of young learners with sight loss who, when undertaking vital education in STEM (science, technology, engineering and math) subjects, lack even a frame of reference for what a graph or chart is in the first place – let alone the data that it relays.

The complex, data-rich and highly visual nature of such models often renders them refractive to the assistive technology, such as screen readers and electronic braille displays, that visually impaired students might use in the classroom.

Often, the only option available is the unsustainable and independence-compromising practice of receiving an in-depth verbal description of the data from teachers or sighted guides.

Over the past couple of months, Benetech – a non-profit social enterprise that focuses on converting books into accessible formats for individuals with print disabilities via its internationally renowned Bookshare platform, has been collaborating with Schmidt Futures to solve this fundamental global issue of educational inequity.

Schmidt Futures is the philanthropic initiative founded by former Google CEO Eric Schmidt and his wife Wendy dedicated to bringing the power of networks and advanced technology to bear in tackling areas of unmet need across science and society.

Their joint initiative builds on ongoing work undertaken by Benetech in conjunction with General Motors using AI to identify, extract and convert over 8 million mathematical equations in Bookshare into accessible formats.

Next level

The latest project addresses a whole new level of complexity as it focuses on building a large dataset that will be used to train AI on how to identify graphs and charts in textbooks and accurately describe them to a learner with a vision impairment or print disability.

With thousands of potential variables at play – the challenge is immense and additionally involves the running of a data science competition to establish best practices in the use of AI models to automate the classification and remediation of graphs and charts.

“Twenty years ago, the accessibility of text was the main challenge and optical character recognition was just beginning to come into its own where you can scan a page and convert it into readable text for screen reading software,” explains Brad Turner Benetech’s VP and GM, Global Education and Literacy.

“Today, the new frontier is STEM, it’s math, it’s charts, graphs and physics. As of now, we’ve got conversion of math equations up to about 99% accuracy and we’re excited to be kicking off our project with Schmidt Futures exploring datasets to describe charts and graphs.”

Educational equality

The challenge is one with global ramifications as Turner outlines:

“Currently, in the global south across huge swathes of Africa, south Asia and Latin America – visually impaired students beyond the age of around 10 or 11 are simply not allowed to study STEM subjects because the content is not accessible.

“It’s not because the student isn’t interested or bright. The content is just not there,” says Turner.

Nonetheless, poor access to STEM education is anything but purely a third-world issue.

Data from the “Women, Minorities, and Persons with Disabilities in Science and Engineering” report published in April 2021 by the National Center for Science and Engineering Statistics at the National Science Foundation reveals that the overall percentage of scientists in the U.S. with one or more visible or invisible disabilities has grown at a disappointingly pedestrian rate from just 6% in 1999 to 9% in 2019.

Additionally, scientists and engineers with disabilities were found to have a higher unemployment rate than those without disabilities and a higher level of unemployment than the overall U.S. unemployment rate in 2019.

“As a society, we often make the mistake of viewing STEM as specialized and something that only a very small subset of students is interested in – when it’s increasingly becoming a core skill area that every student should have a broad familiarity and capability around,” says Kumar Garg, Vice President of Partnerships at Schmidt Futures.

“There seem to be so many different ways that students get coached out of these learning pathways. Education is cumulative and if a student gets knocked off their trajectory of building the core foundational concepts, it becomes really tough for them to try and catch up later,” he explains.

“It's a huge equity challenge if we're not thinking about these access gaps at key points in people's trajectory. These are the rapidly growing areas students are interested in and we should be doing everything we can to actively encourage them and offer support.”

According to Turner, it’s not just the individual student who loses out.

“It’s immeasurable the disservice we are doing to students with a vision impairment or print disability who could have grown up to be the next Galileo, Albert Einstein or Louis Pasteur but have just been steered away from science because the educational content is not available to them. We are depriving wider society of their skills and talents too.

“It’s unjust to rob young people of their true calling in life just because they might read differently whether it’s with their eyes, their ears or their fingers.”

Following this thought process, it is difficult not to reflect on the talent that may have slipped through the cracks already but, perhaps, it is those inequitable educational experiences we still have today that underscore the need to strive for a brighter tomorrow.

Follow me on LinkedIn