The reading achievement gap is actually an opportunity gap that AI can close.

Custom-built AI can ensure objectivity that learners value in evaluation and intervention.

Economical, scalable AI removes the dependency between expensive resources and student performance.

AI enables the individualization of general and special education.

During this Black History Month, let us attend for a moment to the reading achievement gap, as it affects all of us regardless of race. Here's why. Lack of literacy is linked to some of the biggest and costliest problems in society: spiralling special education spending, school dropouts, juvenile delinquency, incarceration, poverty, and mental health (NSBA, 2019; Vacca, 2008; Vacca, 2004; Nelson & Gregg, 2012). We all pay for these problems, at the very least in taxes.

We Spend Over $120 Billion Yearly on Special Education

Reading difficulty, or dyslexia, is classified as a learning disability in special education. Taking up about a third of special ed, dyslexia is the biggest category. As a nation, we spend over $120 billion a year on special ed (NCLD, 2023). The bulk of it comes from our local school taxes.

What is the yield on our yearly investment? Students with dyslexia still read below grade level. They drop out of high school at more than twice the typical rate (NSBA, 2019). Only 5 percent of them go to college, compared to 60 percent of their typical peers (Anthony, 2021). This impacts lifetime earnings, as their peers with college degrees earn about $1 million more than high school diploma holders.

70 Percent of Students Are Not Reading-Proficient

But the literacy problem is not confined to the 20 percent who have dyslexia (Yale Center for Dyslexia, 2022). According to the Nation's Report Card, around 70 percent of students now fail to meet reading standards in some states, including New York and Georgia. While approximately 40 percent of white readers are proficient, less than 20 percent of Black readers are (NCES, 2025).

The Literacy Link: Juvenile Delinquency and Incarceration

In the field of criminal justice, poor literacy skills are recognized factors of crime (Vacca, 2008). The Department of Justice associates juvenile delinquency with low reading ability (Brunner, 1993). African American youths are overrepresented in this system. Half of prison inmates cannot read or write. The National Institute for Literacy reports that nearly half of adults with the lowest literacy skills live in poverty, most with no full- or part-time employment.

The words of a 68-year-old inmate at the California Correctional Institution capture the human and societal cost of these systemic problems. David wrote to our firm when he heard a news report about Dysolve, our AI expert system that resolves dyslexia. He had been incarcerated for eight years.

"Like many inmates here in prison, I didn't get my diploma from high school do [sic] to my learning disability of dyslexia...The one quote I have heard many times [at school] and can't stand hearing is 'Not trying hard enough.' Plus all of the other quotes [being labeled lazy, dumb, careless, immature] are really hard on a kid's self-esteem."

When it comes to literacy, every aspect of a citizen's life is affected, from the academic, legal, and economic to the psycho-social.

AI Drives Transformational Change for Endemic Problems

But during this particular Black History Month, we can go beyond mere reminders of these endemic, systemic problems. We are at the beginning of an AI revolution that is transforming many fields all at once. The reading problem is not immune.

In my previous post, "AI Beat Reading Interventions in Clinical Trial," I mentioned that our expert system managed to break a long-standing barrier: clearing reading difficulty in students after third grade. With other interventions, reading research shows that older struggling readers will continue to struggle throughout school regardless (Elliott & Grigorenko, 2014).

The Dysolve clinical trial, a randomized controlled trial, is noteworthy because participation was primarily from the group that has often been left out of screening, funding, and research. Approximately 80 percent of the trial's participants were minority from low-income communities. They were in grades 3 through 8, performing at the 10th percentile on average in state and school standardized reading assessments pre-intervention (May & Van Horne, 2025).

Implications of the First AI Educational Intervention Trial

That the trial produced a positive effect in this population has several implications:

The debate over poor vs. dyslexic readers becomes moot. This computer system is focused only on clearing language processing difficulties that hinder reading development, not on how they came to be. Previously, some quarters wanted to differentiate between reading difficulty caused by poor learning environments vs. genetics or neurology. A neuropsychological evaluation to get a diagnosis costs $5K to 10K per person. A dyslexia diagnosis thus becomes an economic, not a clinical, issue. To the AI system, however, the label does not matter.

AI-powered intervention is nonjudgmental. This synthetic expert cannot see the student. It cannot use the multitude of clues from physical appearance, demeanor, etc., to form impressions about this person that may bias evaluation results. For this reason, some struggling learners, especially older ones, prefer dealing with a computer program to human specialists.

AI severs the relationship between results and resources. Previously, low-income districts showed poorer results due to scarce resources for high-quality teachers, teacher training, and instructional materials. The point of well-designed AI is to solve the problem of scalability, cost, and outcomes. The Dysolve trial occurred during the COVID pandemic, when schools were scrambling to find any teacher, or any paraprofessional, to supervise classrooms—similar to what is still happening in struggling districts presently. The AI technology is designed to work even under these trying circumstances, yielding a positive impact in the trial.

Challenging settings require an autonomous plug-and-play program. The supervising adult's role is simply to put the child in front of a computer or tablet and let the AI expert do its work. Dysolve AI presents interactive game activities tailored to that student's particular learning problems, continuously on demand. What underfunded districts cannot afford is paying $10K to train each teacher and $10K to 20K for each special ed pupil a year indefinitely due to ineffective methods. An efficacious AI solution is thus also an economic solution.

Individualization of Education Levels the Playing Field

This AI innovation is a precursor to the individualization of education. Each student gets an individualized program specific to their strengths and weaknesses. We are beginning to see its impact on struggling learners who are motivated to seize the opportunity. When they do, they are empowered to change their own developmental trajectory.

Jesseme Lynch from North Carolina explained how his reading difficulty affected his self-esteem in third grade: "It kind of made me feel disappointed in myself because I couldn't read the words like everybody else." Three months later, his attitude changed along with academic improvement: "After Dysolve, I can tell what the words meant, and I could get good scores."

Aaliyah Williams struggled to keep up with her classmates in learning and understanding subject matter in fifth grade: "I had a hard time comprehending books and had a significantly lower reading level than where I should have been." After the AI intervention, Aaliyah read on-grade. By her senior year of high school, she took a college-level English course and subsequently thrived in college.

Cases such as Jesseme's and Aaliyah's remind us that the reading achievement gap is really an opportunity gap. When we are able to give these children the opportunity to excel, they often do.

Now that's worth celebrating during this Black History Month.

My next post will be about an 88-year-old civil rights activist with dyslexia who is using his own positive experience with the AI program to advocate for others. Stay tuned.

National School Boards Association (NSBA). (2019). Good news, bad news on graduation rates. Data on Disabilities.

Vacca, J. S. (2008). Crime can be prevented if schools teach juvenile offenders to read. Children and Youth Services Review, 30(9), 1055-1062. https://doi.org/10.1016/j.childyouth.2008.01.013

Vacca, J. (2004). Educated prisoners are less likely to return to prison. Journal of Correctional Education, 55(4), 297-305.

Nelson, J. M., & Gregg, N. (2012). Depression and anxiety among transitioning adolescents and college students with ADHD, dyslexia, or comorbid ADHD/dyslexia. Journal of Attention Disorders, 16, 244–254.

The Yale Center for Dyslexia and Creativity. (2022). Declaration of rights.

National Center for Learning Disabilities. (2023). IDEA full funding: Why should Congress invest in special education? Policy & Advocacy. https://ncld.org/news/policy-and-advocacy/idea-full-funding-why-should-…

Anthony, N., & Bills, K. (2021). Examining college stop out rates for students with disabilities. Journal of Education & Social Policy, 8(3), 13-18. doi:10.30845/jesp.v8n3p2.

National Center for Education Statistics (NCES). (2023). Students with disabilities. Preprimary, Elementary, and Secondary Education. https://nces.ed.gov/programs/coe/indicator/cgg/students-with-disabiliti…

Brunner, M. S. (1993). Reduced recidivism and increased employment opportunity through research-based reading instruction. Washington, DC: Office of Juvenile Justice and Delinquency Prevention, US Department of Justice. NCJ 141324

Elliott, J. G., & Grigorenko, E. L. (2014). The dyslexia debate. New York: Cambridge.

May, H. & Van Horne, S. (2025). Results from a randomized trial of the Dysolve Program for students with reading difficulties. The Center for Research in Education and Social Policy (CRESP), University of Delaware.

© Psychology Today