Before today's tumult, special education in the U.S. was already in crisis. It's just that now, the general public is paying attention to special ed. Indeed, the public should be engaged, as special ed affects everyone, not just the school staff and students.
Consider how the public may view these stats:
- More than $120 billion of our tax money goes to special ed every year.
- By 2023, approximately 7.5 million students were in special ed nationwide.
- While special ed used to take 20 years to add a million more students, it now takes only 5 years.
- Despite spending twice as much on a special ed vs. general ed student in some states, substantially more of the former drop out of high school and forego college.
Let's review how we got here and how we can get out of this crisis.
Reading Crisis
According to the Nation's Report Card, approximately two thirds of 4th, 8th, and 12th graders fail to meet reading standards. While widely known in school circles, this number is shocking to the layperson.
As reading scores fall even further, long COVID takes on a new meaning. We are only beginning to see the long-term impact of school shutdowns on young children during the critical years of their development.
When asked what percentage of their students have a reading disability or dyslexia, administrators' estimates range widely from under 5% to 20%. Regardless, dyslexia demands our attention as it has the largest share of students in special ed.
And regardless of type or intensity, reading interventions have not been able to get these students in 3rd grade and beyond to read on grade level.
Despite the popularity of some intervention approaches, meta-analytic reviews of efficacy studies repeatedly show the absence of a significant positive effect on broad reading achievement for this population. These reviews also point out the lack of scientific rigor in many intervention studies as well as the glaring presence of the research-to-practice gap.
Schools are at a loss as to what to do with struggling readers after 3rd grade. By middle school, directors of special ed have tried everything in their toolbox. Researchers know about the Matthew Effect in this area: proficient readers surge ahead while failing ones fall further and further behind.
The layperson may think that schools automatically give additional support to students who fall below the 50th percentile. In reality, even those in the bottom third do not automatically receive special services.
Considering that special ed serves approximately 15% of students (for all disabilities), and yet 20% of all students have dyslexia, that means that a sizable number of the latter are left out. Most are not even diagnosed.
Economic Crisis
Are we spending too much or too little on special ed? The answer is complicated. We will focus on dyslexia, given its significance in special ed.
For other disabilities, an estimated 9% of students transfer out of special ed a year nationally. But students with dyslexia generally continue to need resource-intensive support throughout their school years.
This is because the primary reading intervention used, multisensory structured literacy, is compensatory and merely helps students cope. Their reading disability remains chronic, requiring long-term care.
And this care is costly due to intensive, lengthy teacher training and one-on-one or small-group instruction. Public schools cannot afford to use this method for all their struggling readers. There are simply not enough trained special ed teachers. Post-pandemic, there are not even enough teachers, period.
Not surprisingly, the number of IDEA disputes between families and schools has risen sharply, as have lawsuits. Parents' awareness of this mandate to provide a free, appropriate education to students with disabilities is growing. So is awareness of dyslexia. Hence the increase in legal challenges.
When states try to address the dyslexia problem actively by increasing the number of students getting reading intervention, they end up with an economic problem. For example, Texas now faces a $2.3 billion annual deficit in this area.
Problem and Solution
In short, current methods are not effective and are too expensive to serve everyone in need. What is needed is a method that is
- effective
- inexpensive
- resource-light
- easy to implement
- accessible to all
The solution lies in AI.
AI Solution for Dyslexia
Although it is not usually described in this way, dyslexia is fundamentally a language processing problem. Researchers agree that dyslexia involves the linguistic system in the brain, although some may give more weight on one component such as the phonemic (speech sounds).
Therefore, to solve dyslexia, we need to locate the sources of the processing problems in a person's linguistic system. But this is challenging for these reasons:
- The linguistic system is super complex. The method would require analysis and cross-referencing of billions of data points per person.
- The method has to work super fast due to the speed of language processing in the brain.
- The method has to accommodate the wide diversity and substantial size of the dyslexic population. This involves approximately 10 million children in the U.S.
When the problem is presented in this way, we can see why no human specialist, no matter how well trained or experienced, has been able to correct dyslexia for the past 100 years.
But this task is not beyond the capability of computers.
A Scalable Program
The dyslexia problem has indeed been solved by AI. The dyslexia intervention program in commercial use is powered by autonomous AI. An autonomous AI system makes decisions independently without human input.
This autonomous AI takes a student from signup through screening and intervention automatically without human support needed. It uses a game interface to interact with a student.
When a person plays a game, certain targeted brain processes are activated for analysis. Critical bits of information about these brain processes are contained in the user’s game responses. Which processes ran efficiently, and which ones did not? At what level of efficiency?
This autonomous AI system generates and delivers online games continuously to a person with dyslexia, using previous gameplay data to build the next game.
Dyslexia intervention requires autonomous AI because it has to overcome the three major obstacles described above: complexity, speed, and scalability.
Because of the sheer speed of language processing, an autonomous AI expert system is essential for dyslexia. When a student sits at the computer to start their intervention, the expert system has to create a game, analyze the user’s game responses, cross-reference them with relevant databases, and generate the next game uninterrupted.
AI can do all this in real time—and for millions of students one on one.
Cost Savings
Autonomous AI overcomes longstanding problems in special ed, including chronic reading difficulties, lack of access to services, and exorbitant costs.
Traditional dyslexia interventions cost more than $10-20K for each student a year in most states. Because these compensatory methods are still needed for the rest of a student’s school years, their total cost is over $100-200K per pupil.
In contrast, a scalable AI program costs only 1-10% of this amount, when the dyslexia is corrected in 1-2 years. These huge savings benefit all students, when districts no longer have to make wrenching decisions over budget cuts.
Future
Doing business as usual in special ed is no longer sustainable. It is time to pause and reflect on the path forward.
Innovative AI technologies can resolve the endemic problems plaguing the special ed wing of public schools: ballooning costs, poor student outcomes, teacher shortages, and legal challenges.
Often, school systems are slow to adopt new technologies as change inevitably involves more work. Not anymore. The purpose of automation is to reduce workload. The new generation of technologies is designed to make implementation effortless.
About the author
Coral PS Hoh, PhD, is a clinical linguist and architect of Dysolve® AI. She is the founder and CEO of EduNational LLC, the maker of Dysolve® AI.
