A recent article in EdWeek was headlined, “What Educators Need to Know About Dyslexia—and Why It’s Not Something to ‘Fix.’” As an inventor of autonomous AI, I explain below why technological advances actually can resolve previously intractable problems, and why the case of dyslexia shows how artifical intelligence can change how we view and treat such conditions.
Dyslexia is a language-based neurodevelopmental condition that causes difficulties with word recognition, spelling, and reading comprehension. It affects 1 in 5 people and is the most-cited condition qualifying for special education (NCES, 2024). Special education programs overall cost U.S. taxpayers over $120 billion a year (NCLD, 2023). EdWeek‘s recent article asserts that dyslexia cannot be cured (or “fixed”) although interventions can help students compensate and manage their reading difficulty. The high cost of dyslexia is partly due to recurring expenses from interventions that can only help struggling readers cope but not correct this neurodevelopmental disorder.
A Logical Approach
What is indisputable from 5 decades of research is that dyslexia is a language processing difficulty (Elliott & Grigorenko, 2014). Logically, therefore, a dyslexia solution requires these:
- Expertise in natural language
- Expertise in language processing in exceptional brains
- A method to identify and correct language processing difficulties in individual brains
1. Expertise in natural language. Linguistics, the scientific study of natural language, comprises many areas of specialization, such as Phonology (sounds), Semantics (meaning) and Syntax (sentences). Researchers with PhDs in Linguistics typically specialize in no more than 1 or 2 areas. Yet any area of the linguistic system in the brain may be affected when there is a language processing issue.
2. Expertise in language processing in exceptional brains. Linguistics provides knowledge of natural language as an abstract system. To solve dyslexia, we also need to know how the brain processes language. Moreover, we need to know how exceptional (specifically dyslexic) brains process language v. typical brains.
3. A method to identify and correct language processing difficulties in individual brains. The problem with the dyslexia concept is that it does not have an operational definition that researchers can all agree to (Elliott & Grigorenko, 2014). Precision is key to solving this condition.
First, we need to define “difficulties” in a precise, measurable way. We may define them as “inefficiencies,” which can then be measured against a target, such as full efficiency (100% efficiency). Next, we need to devise a method that can measure inefficiencies in individuals’ brain processing in a convenient, non-invasive way.
Technical Requirements
This new method requires these capabilities:
- Analyze billions of datapoints per person, given the complexity of the linguistic system and language processing in the brain.
- Operate at the rapid speed of language processes, which often occur in hundreds of milliseconds.
- Measure sub-second changes in language processing to track improvements.
- Perform (a)-(c) at scale for the diversity of individuals with different scope and severity of processing inefficiencies.
These technical requirements are beyond any human specialist, no matter how well trained. What is needed is an autonomous computer system that can perform the above operations while meeting the exacting demands of capacity, precision and speed of the dyslexia problem. This autonomous system has to serve as the expert, evaluating and retraining each person to increase the efficiency of language processes in their brain.
To activate these processes in the brain, this expert system may generate an activity, provide the input and collect the user’s output to analyze. This activity can be packaged as a language-based, online game, which then serves as the interface between the user and the AI system. For example, the input can be a word flashed on the screen and the user’s output the oral reading of that word. The AI expert then determines whether the output is correct. If not, the system may analyze the user’s incorrect output to look for possible sources of error. The results of this error analysis help the AI expert decide what activity to generate next.
This kind of individualized intervention created in real time is now possible, with the advent of cloud computing, dynamically scalable technology, and interactive gaming. AI thus provides a new view of dyslexia, now defined computationally. The AI method involves input-output functions. Instead of the physical brain, it deals with the functional brain in a convenient, inexpensive, non-invasive manner. AI provides a quicker solution today, given that the ability to identify dyslexic traits from scans of physical brains may not be possible ever (Ozernov-Palchik & Gaab, 2016).
That is to say, advances in AI can change how we view and treat a brain condition like dyslexia. Old assumptions about neurobiological disorders being ‘lifelong’ may need to be revisited (Huebeck, 2025). The brain’s ability to reorganize, neuroplasticity, is only just beginning to be understood presently.
Cases
AI technology, guided by the requisite knowledge base, has already corrected dyslexia in individuals in the U.S.. Last year, a dozen of them described in regional and national media the end of their reading difficulty after receiving AI-driven intervention for 1-2 years. NBC Nightly News featured one such case last December (Ellis, 2024). Prior to the AI intervention, many of them struggled for years with reading despite using various other interventions. After the AI intervention, they performed like their typical peers academically. For years, these students were told that they could at best merely “manage” their dyslexia and learn to “cope” with it. That they would continue to struggle with reading and language classes. That they could never hope to read on grade level or excel in class.
New AI technology changed all that.
