AI-Powered Solution for Learning Disabilities

Author(s): Scott Douglas Jacobsen

Publication (Outlet/Website): The Good Men Project

Publication Date (yyyy/mm/dd): 2025/02/13

Dr. Coral Hoh is a clinical linguist and co-founder of Dysolve. This AI-driven platform supports students with learning disabilities, particularly dyslexia. Through innovative, game-based assessments and personalized educational plans, Dysolve has helped hundreds of students achieve grade-level reading within 1–2 years at a fraction of the cost of traditional special education programs. Dr. Hoh combines linguistics, cognitive science, and AI expertise to advance accessible, effective learning solutions while addressing ethical and privacy considerations in education technology. Hoh shared insights on their AI tool, designed to address learning disabilities like dyslexia by correcting language processing deficits. Unlike traditional methods that compensate for dyslexia, Dysolve aims to resolve it by achieving 90–100% student efficiency. Dr. Hoh emphasized the emotional and mental health benefits for children who overcome these challenges, noting significant transformations in behaviour and academic performance. With a focus on privacy and custom-built AI, Dysolve’s approach highlights collaboration across linguistics, mathematics, and software engineering to address this issue holistically. Dr. Hoh announced their goal of ending dyslexia entirely.

Scott Douglas Jacobsen: So today, we’re here with Dr. Coral Hoh, a clinical linguist and co-founder of Dysolve. Dysolve is an AI tool designed to assist teachers in personalizing educational experiences for students with learning disabilities. The assessment focuses directly on the students. You’ve helped students achieve grade-level reading, even those with dyslexia. How does this improve cost efficiency compared to various special education programs at the state level? And, personally, how did you become interested in linguistics, education, and related fields? 

Dr. Coral Hoh: How did I end up in linguistics? I come from a place where people are multilingual. Language is something we grew up with—I’m from Malaysia. When I went to college, I joined a program in linguistics, though it wasn’t explicitly called that. It was called English, but it was linguistics all the way through. Early on, I realized this was what I loved because it’s like being a forensic expert. You can’t see the brain directly, but by examining obvious clues—what a person says, how they respond, what’s missing, or what’s incorrect—you can infer what’s happening inside.

I began developing this method during my bachelor’s program. I continued refining it when I entered a Ph.D. program in linguistics at the University of Delaware. By then, cognitive science and AI were gaining popularity. Even back then, we were already discussing neural networks. After earning my Ph.D., I moved to upstate New York and began working with local communities. I discovered that many families were struggling with language learning problems.

I love working with children, but it became an emotional journey as I realized how big of a crisis this was for many families. Disabilities like dyslexia are common, and they can tear families apart. I ended up working with single mothers frequently, which is especially challenging for young moms when boys grow quickly. For example, a 10-year-old boy, based on research, can experience a growth spurt that leaves him taller than his mother within a year. At that point, it becomes difficult for a mother to command authority when she has to look up to her son.

This made me see the seriousness of the issue. When I delved into the research on dyslexia, I found it vague and inconsistent—it wasn’t operationalized in a way that could provide clear, measurable solutions. Even today, the definition of dyslexia lacks concrete clarity. I realized that even if I could help one or ten individuals, millions were affected. The only scalable solution was AI.

That’s why we built this AI program—to evaluate and determine the severity of language processing problems, which is what dyslexia is at its core. People use the term “dyslexia,” but at a fundamental level, we need to focus on the language processes in the brain.

When you do that, you can measure it for each person. Then, you set about correcting. That’s what Dysolve AI, the system we built, does. So it’s a combination of different fields working together.

And if you talk about my background, I had a foundation in linguistics, but it wasn’t enough because I didn’t do clinical linguistics or learn about the brain. Then, there’s the challenge of translating a linguistic model into something that can be coded. So, you have to delve into mathematical modelling as well. Fortunately, I spent most of my career collaborating with mathematicians and software engineers.

Jacobsen: So, how you were able to manage the translation?

Hoh: Yes. Outside of what we do, people often talk about AI requiring domain experts from various fields to solve problems. Then, you also need translators to bridge those fields. In our case, I handled that translation myself. I designed the system and am also the architect of the program.

Jacobsen: I see. You’ve provided your background and explained your reasons for getting into this work. You’ve also highlighted the humane aspect of addressing difficult social contexts, particularly for families and single mothers. What about defining dyslexia in concrete terms, as you mentioned earlier? The idea is to parse dyslexia into its relevant empirical parts, then code that into Dysolve’s architecture and use it to help individuals with this particular issue. How does that work?

Hoh: You summarized it. Discussing dyslexia as a language processing problem is about identifying what isn’t being processed correctly. Dyslexia often involves deficits in phonological processing, which is how we process speech sounds—phonemes, single sounds, groups of sounds, or phonological patterns.

Jacobsen: Is this well-established in the research?

Hoh: Over 40 years of research have shown that phonological processing is a major predictor of dyslexia. People with dyslexia often struggle with processing sounds, but that’s only one component of language. There are also larger units, like words, sentences, and semantics—the meaning of language. Unfortunately, these aspects tend to be ignored because sounds are finite, concrete, and easier to study than language’s more complex and abstract components.

Jacobsen: So, has the research largely focused on the phonological aspect?

Hoh: Exactly. While it’s valuable, it’s not enough. Dyslexia can affect any part of language processing. Addressing just the sound component won’t solve the problem fully.

Jacobsen: Are there case studies of individuals with pervasive dyslexia who face challenges across all factors associated with the condition? Do such individuals take longer to benefit from the program due to the ubiquity of their deficits?

Hoh: Yes, that’s common. Some individuals have pervasive dyslexia where all factors—phonological, semantic, syntactic, and more—are affected. It presents an overwhelming hurdle, and while the program helps, it takes longer for these cases because their deficits are so widespread.

So, especially early on, the students who came to us had severe problems. That’s why their parents were actively looking for a solution. It’s severe enough that they couldn’t just continue with other methods—they had to address this issue. Many of these students had significant language processing issues.

Since we actually count and measure their progress, we aim to get them to 90–100% efficiency. This is measurable because the program is game-based, and we track their performance accurately in each task. We assess accuracy along with the speed of response and delivery, which together define efficiency.

Many of these students started at 20% efficiency—very serious cases. They often had multiple conditions, such as ADHD, autism, and psychological issues, all overlapping. Addressing these challenges takes longer, and typically, if they’re working regularly each week with occasional breaks, it takes about two calendar years. For others, without such severe conditions, it generally takes about one calendar year.

When multiple language functions are affected, dyslexia can overlap with dysgraphia. Dyslexia involves difficulties with reading, while dysgraphia affects writing, particularly sentence composition. Rather than categorizing them strictly as “dyslexia” or “dysgraphia,” we see it as a continuum. That’s why focusing solely on sounds isn’t sufficient—you must also address sentences.

Jacobsen: What about data collection issues? A big cultural topic around AI right now is ethics. I recently spoke with someone working on semantics and AI, and part of the discussion involved how AI relies on large amounts of data—both real and synthetic. For data gathered from children, are there concerns about privacy and security? Or is the data you’re gathering less sensitive and unlikely to be misused?

Hoh: Yes, that always comes up when we mention AI—data privacy and security. The first thing people need to understand is that there are different types of AI. Ours wasn’t built by simply mining everything available. It’s much more selective and custom-built.

We didn’t use generic tools to develop this. The entire system, from top to bottom, was specifically designed for this purpose. Users build their own siloed database, and no one else has access to it. We don’t share data with third parties, so it’s different—the data are not distributed throughout the program system or beyond.

So, it’s different. I don’t particularly like the term “siloed AI” because some people see it as a negative concept. But in this case, it is compartmentalized—it’s not shared with third parties or anyone else.

Jacobsen: That’s a safe and reasonable way to approach it. The future of many technologies involves specific, actionable tasks. What about timelines? You mentioned two years for cases with multiple combined deficits and about one year for less severe cases. Are you seeing improvements from 20% to 90% efficiency as a linear progression, or does it tend to be more of a slow build followed by a hockey stick curve?

Hoh: It’s more of a hockey stick. If you look at the details, the graph fluctuates. When students get close to 90% efficiency in one type of task, Dysolve AI doesn’t stop there. It shifts focus to another area where they still have challenges, which can cause a temporary drop in performance.

Initially, there were ups and downs as the system worked through the various problem areas. This makes sense because the AI is methodically addressing all the deficits. Then, at some point, after the major areas have been resolved, the brain “clicks.”

When that happens, many pathways clear simultaneously. This often results in a sudden transformation that everyone notices—parents, teachers, and even the children themselves. It can happen in as little as a week. A child who was previously moody, resistant to work, or withdrawn becomes optimistic and cheerful and starts volunteering to read. It’s like seeing a completely different person.

Jacobsen: Do children experience improvements in mental health or well-being when they no longer face pervasive academic struggles?

Hoh: Absolutely, and that’s a crucial aspect of this work. It’s one of the reasons we wrote a book called Dyslexia Dissolved. The book is primarily for parents and includes 10 different cases to illustrate the various facets of dyslexia and related struggles.

For example, we talk about a boy who was bullied extensively. He was punched in the stomach at school because other kids realized he couldn’t read. One day, he fought back, but only his reaction was noticed, so he was labelled a bully. Depression is another common issue.

There’s also a case of a girl who was functionally mute. Her challenges weren’t just with reading but also auditory processing—being able to comprehend and organize words into sentences. She became numb, emotionally withdrawn, and stopped interacting with her family. She even refused to hug them and completely isolated herself.

If this had persisted into adulthood, it could have been devastating. Fortunately, we were able to help her. She graduated from college and even became a school leader in high school. That’s the power of addressing these issues holistically—academically, emotionally, and socially.

Even if it takes two years—or in some cases, less than one year—the transformation is remarkable. The whole person changes. Mental health improves, depression lessens, and we’ve even seen children with hypertension suffer from this issue. It’s overwhelming.

People often think of this as just a reading problem. But even when the book is closed, the problem doesn’t go away for them because it’s about language processing. If you consider how much of your waking hours are spent processing language—it’s constant—all the time.

This is a big issue. With about 20% of the population affected, it’s a huge one. 

Jacobsen: Do you see yourself collaborating with other teams or companies specializing in different areas of learning disabilities?

Hoh: That’s a good question. I see many ed-tech companies, and we’ve had early-stage discussions about potential collaborations. However, most of them focus entirely on education, while we also address the aspect of brain disorders. They typically create programs aimed at compensating for disabilities.

For over 100 years, people have been trying to compensate for dyslexia, this language processing problem. But what we’re doing is correcting it—ending dyslexia. This year, we’re excited to announce that this is the end of dyslexia. People need to use the program.

We’ve documented these cases in our book; those students are fine now. Just last year, a dozen former students came forward and shared that their dyslexia had been resolved. They’ve experienced significant transformations.

Jacobsen: Dr. Hoh, thank you for your time and this insightful discussion. I appreciate the opportunity.

Hoh: My pleasure. 

Jacobsen: Nice to meet you.

Last updated May 3, 2025. These terms govern all In Sight Publishing content—past, present, and future—and supersede any prior notices. In Sight Publishing by Scott Douglas Jacobsen is licensed under a Creative Commons BY‑NC‑ND 4.0; © In Sight Publishing by Scott Douglas Jacobsen 2012–Present. All trademarks, performances, databases & branding are owned by their rights holders; no use without permission. Unauthorized copying, modification, framing or public communication is prohibited. External links are not endorsed. Cookies & tracking require consent, and data processing complies with PIPEDA & GDPR; no data from children < 13 (COPPA). Content meets WCAG 2.1 AA under the Accessible Canada Act & is preserved in open archival formats with backups. Excerpts & links require full credit & hyperlink; limited quoting under fair-dealing & fair-use. All content is informational; no liability for errors or omissions: Feedback welcome, and verified errors corrected promptly. For permissions or DMCA notices, email: scott.jacobsen2025@gmail.com. Site use is governed by BC laws; content is “as‑is,” liability limited, users indemnify us; moral, performers’ & database sui generis rights reserved.