October 4, 2023
Josh Clark, The Landmark School
In February of 2020, reports of Chinese schools employing unheard of mitigation strategies to combat a mysterious and growing virus caught my attention. I watched video clips of teachers clad in medical armor taking students’ temperatures before spraying the bottoms of their shoes with bleach. It did not take long for these defenses to fail, and seemingly overnight Chinese schools were closed, and students and teachers exiled online.
Everything about it seemed impossible. I took comfort in my certainty that this could never happen in the United States. Yet I could not help but wonder: What if? What if this invisible menace eluded our defenses? What if the world as we knew it suddenly changed? Personal matters aside, how could we possibly teach dyslexic learners online? I became obsessed. I read every news story building an arsenal of information and ideas. I was a digital doomsdayer hoarding not toilet paper but Google Meet tutorials.
On February 26, 2020, I sent an email about the ‘coronavirus’ to a listserv of approximately 200 administrators at schools for students with dyslexia, asking, “What if what happened in China happens to us?” I qualified the question by highlighting its absurdity, assuring everyone (and myself) that we would soon laugh about this, cozied up to the bar at the next conference. I felt like Chicken Little screaming that the sky was falling.
The first time I used OpenAI’s ChatGPT-4, I felt the sky starting to crumble again.
The same cocktail of anxiety and adrenaline that sent me down online rabbit holes about air filtration and ‘synchronous online learning’ is again monopolizing my attention. Much like March of 2020, I am wondering how we can rapidly adapt to unprecedented and unpredictable change. There are many important differences between the impacts of Covid-19 on education and the potential of generative AI, though they may not be what you think. There is a chance they both could kill us and just like the early days of lockdown, there is a lot we do not know, even as it begins to transform our lives. Unlike our response to Covid, however, AI-inspired changes will not be temporary.
As leaders of schools for students with dyslexia, the implications of this technology on our students and our programs are vast. Reading and dyslexia researcher Dr. Mark Seidenberg urged us to act upon all we know about “Language at the Speed of Sight.” What will it mean for our students, our instruction, and our institutions when language is at the speed of WIFI?
As I share some of my early thoughts and questions, two assumptions inform my thinking:
What Is Generative AI?
Generative AI is an emerging technology that uses complex, predictive algorithms called large language models to generate new, realistic content including text, video and images. Put simply, it analyzes massive amounts of data from the Internet to create content, solve problems, and answer questions in a way that feels “human,” and in some cases, exceeds our own capabilities. And it does so in seconds.
Presently, the best-known example is ChatGPT. It is a free website (with a paid upgrade option) that can generate any form of text about anything in any language and at any level. While it is far from perfect, even if no improvements were made (and they will be), in its current form the technology has existential consequences for how we think about teaching, learning, and working – not to mention philosophical and ethical questions about originality, creativity, etc.
Why Does This Matter?
An understandable response to this new technology is a simple sigh, as we relegate it to the wasteland of past breakthroughs guaranteed to “revolutionize” teaching. Unlike your stupid SMART board, though, even in its earliest of iterations, this technology will impact all aspects of how we teach, work, learn, and live.
It is already predicting complex protein structures, helping doctors become more efficient in diagnosing patients, and built a functioning website based on a picture of a description sketched on a napkin. Even more amazingly, it was not designed to do any of these tasks. It learned to do them. It is the technology’s ability to generalize its application that has many of its own developers describing its creation as a ‘summoning’ more than an invention. In many ways, even the engineers building it do not fully understand how it works or what it can do. All we know is that it is happening very quickly and is just beginning.
Microsoft recently shared its AI-powered Co-Pilot, and its underlying technology will likely change the way we do work, eliminating many repeatable and routine tasks. It is not hyperbolic to think that in the near future we will all have AI personal assistants managing our calendars, writing student reports, summarizing recordings of our classes, generating lesson plans and assessments, and creating individualized content for students. Not to mention personalizing our nutrition goals and then generating corresponding recipes for tonight’s dinner based on pictures of our fridges and pantries.
AI will not just make us skinnier and more productive; there is a darker side as well. Microsoft’s Bing search engine is now powered by ChatGPT, and when you ask a question, it provides a narrative response as opposed to a list of websites. This is highly disruptive to search’s traditional revenue stream from advertisements. Keep in mind that those ads are strategically curated for you based on your search and browser histories. Those same histories also influence the kinds of stories you see on social media and the videos you watch on YouTube. Those manipulative algorithms might also be why your Uncle John transformed from your favorite relative into a QAnon fanatic wearing an aluminum foil hat to Thanksgiving dinner. This kind of technology will soon seem primitive if AI is used to manipulate our thinking.
Over the next decade, we are likely to experience industrial-revolution-size efficiency gains and potentially destabilizing online misinformation and manipulation. These forces will change the world we are preparing students to enter. We need to start thinking about this technology now because we are already behind.
Won’t Students Just Use ChatGPT to Cheat?
Yes, but only if we continue to assess students in the same outdated ways. The Siena School produced a fantastic webinar on this topic. Soon ChatGPT will be as expected and ubiquitous as spell check. Spell check empowers poor spellers by preventing their difficulties from impacting their ability to effectively communicate. In the near-term the same will be true for individuals who have difficulty organizing their ideas through written expression. Everyone will become an effective writer.
AI Could (SHOULD) Change the Ways We Measure Potential
ChatGPT4 can outperform the vast majority of people on some of our most consequential standardized exams. It scored 700/800 on the SAT math section, in the top 10% of the bar exam, in the 85% percentile of the LSAT, and a 4 and 5 on the AP calculus and AP biology exams. It even aced the Sommelier exam. Interestingly, it did not fare as well on the AP language exams which require higher levels of creativity and interpretation.
In a world where a computer can outperform many humans on these exams, how can they possibly be an accurate prediction of who will be successful? In fact, they would seem to only predict who has the skills that can be better performed by computers. Dyslexic learners stand to be the beneficiaries of this shift in what is important and how we measure it.
Further, any school that continues to build its curriculum, instruction, and culture around these measures will soon find themselves preparing students for an anachronistic world.
AI Will Change Instruction
AI is likely to have enormous impacts on instruction, especially in schools that serve students with dyslexia. I believe this transformation will begin by giving teachers back time through content generation before eventually changing the role and skill set of the entire profession.
AI can already generate word lists, controlled texts, and adequate lesson plans. It can produce lists of real and nonsense words with certain features – vowel teams, syllable types, spelling patterns, prefixes, etc. – and then incorporate the words into a text about any topic of your choosing. Imagine being able to provide every student with individualized, decodable passages of varying levels of syntactic complexities about basketball, space, race cars, Pokémon, or whatever other topic they find interesting. And generating it all in seconds not hours and days.
Khan Academy is already experimenting with AI-powered personal online tutors. With access to hundreds of thousands of examples of students’ work, AI will learn from student errors and predict and introduce the most efficient interventions. In the near-term, imagine AI reviewing students’ writing samples and providing feedback in real-time. Further, imagine students writing in response to a differentiated text about a subject they find interesting and getting instant feedback and opportunity to improve. Might we discover that for some students who struggle with writing and comprehension it is partially because we ask them to read and write about subjects for which they have no background knowledge or interest?
The possibilities and consequences cascade from there. With enough data to inform and refine its predictions, it is highly likely AI will soon be able to teach dyslexic students to read. Through the instant individualization of subject matter, complexity, and word choice, combined with an arsenal of error analysis and real-time response to student work, AI will be able to predict what kinds and the amount of exposures students need to advance. Further, with the advances in video and audio-based AI, students could choose to have Spider Man as their Language Arts teacher on Monday and Cinderella on Tuesday. AI does not need to become an OG Fellow, Wilson Certified, or a LindamoodBell tutor; it just needs access to sufficient data and a measurable goal.
Soon, technology may provide every student the individualized, explicit, structured, sequential instruction, and expertise that is presently only available to a privileged few. It will also mean that the cost and time of training teachers will diminish exponentially. A 30:1 student to teacher ratio might finally make sense when every student is given an individualized, responsive curriculum.
This may be an appalling notion to many. How could any child learn more from a computer than a human being? Teachers and personal relationships will continue to play a vital role. Integrating multi-sensory practices, building authentic relationships, designing ways for students to practice collaboration and group problem-solving, and managing endless online distractions will become the focus of teachers’ time and purpose rather than charging through a standardized curriculum built for all but appropriate for none.
Further, in a world where over 50% of all American 4th graders are reading below grade level, what is it we are so afraid of losing?
AI Will Change Schools
The dramatic changes in how schools think about and deliver instruction will be mirrored in how they operate. In a world where all curriculum is individualized, schedules can be built around collaboration and growing social skills rather than only content and ability. Calendars could change as the rate at which students display mastery will determine their advancement more so than the traditional school year. Where learning occurs can also change when the pedagogy and expertise is housed in the cloud and not the classroom.
Grading, individual education plans, student reports and other time-consuming paperwork will be automated and more individualized than ever before. When an administrator wants an update on a student’s progress, AI will simply scan and analyze every assignment, grade, note, report, and email about the child and provide a real-time summary.
With fewer, differently trained teachers able to create better results for more students along with many administrative tasks now automated, the cost of education might finally decline.
Existential Threats in Need of Our Immediate Attention (Or Two of Many Questions That Are Freaking Me Out!)
As a leader of an independent school for students with dyslexia, the potential of AI overwhelms me. When I allow myself the space to consider the future, two key questions inspire equal parts excitement and angst:
How Do We Prepare Students for What Comes Next?
In 2005, early into my teaching career, I read Thomas Friedman’s recently published book, The World is Flat.Its predictions of how the Internet and globalization were and would transform the world order became the keynote topic of every educational conference. Again and again we were reminded that we were preparing students for jobs that did not yet exist. It was an inspiring and daunting call to action that mostly proved credible. Eighteen years later, there are all kinds of professions we could not then fathom, such as an app developer and search engine optimization specialist.
As we raced to incorporate technology into our instruction, though, we remained tethered to a traditional scope and sequence of skills. Memorizing state capitals may no longer be important, but you still had to practice writing a five-paragraph essay about their advantageous geography. Even if you wanted to become a lawyer, you still had to take advanced algebra in high school because those complex problems would be on the test that determined your next options. If you struggled with reading fluency or written expression, college may not be for you.
AI will have similar and likely broader implications than the Internet and at a much faster rate. What will it mean for schools when some of our core focuses are automated? How will we keep students engaged in tasks that they know they will never have to perform? My instinct is to argue that reading and writing about ideas will always be an essential part of being human and our ability to function in and contribute to society. And yet, I also have no idea how to grow lettuce, butcher a cow, or purify water. If technology cast-out these essential elements of survival from a well-rounded education, what else could become superfluous?
For students with learning differences, how will we balance remediation with empowerment? Will we deny our students a technology that is more efficient and effective to improve a craft that may not be required of them?
I think about Dr. Maryanne Wolf’s last book, Reader, Come Home. In it, Dr. Wolf laments what the rise of technology and digital scrolling, as opposed to deep reading, could mean for our cognitive abilities. She explains the generative impacts that focused, sustained reading has on our ability to employ empathy, think critically, and problem solve. Deep reading does not make us smarter through its content alone but also through its cognitive exercises. With the rise of ChatGPT I can only imagine her sequel: “Writer, Go to Your Room! You are Grounded!”
I wonder if how we think about literacy as a form of understanding and writing as a method of communication will change with AI. The rise of misinformation is already disrupting our society, and soon it will be ubiquitous and far more sophisticated than outlandish conspiracy theories. In a world where any information can be found instantly and any content created in seconds, will analyzing and critically critiquing information become more important than generating it? Will schools no longer be about learning and producing, but instead, understanding, analyzing, and concluding? We could argue there is no real pedagogical difference between the two. However, how many of us continue to fall for phishing emails or take a social media post as “truth”?
If our job is to prepare students for what comes next, we need to start reckoning with the new next.
What Happens When the Playing Field is Leveled?
What will it look like for independent schools, especially independent LD schools, when the playing field is finally leveled? For many of us, our value proposition is rooted in our ability to offer an individualized education delivered by experts in their fields of study. What will be the competitive advantage of independent schools when AI empowers public schools to easily achieve this standard?
As independent schools begin to consider what AI will mean for education, we must simultaneously consider how these changes will impact our market differentiators and business models. For instance, a traditional independent school justifying a high tuition because 65% of their faculty have masters in their content areas and an average SAT score of 1500, soon may not matter. Conversely, an LD independent school claiming it can teach your child in ways other public and private institutions could be just as hollow.
When I consider the future of AI and independent schools, I center on the idea of authenticity. The company Replika provides users with an animated AI-powered companion. You can video call your Replica 24 hours a day, and it not only listens and responds, but it also remembers your past conversations so it can personalize your relationship, deepening the sense of connection. This kind of technology can be life-changing for older individuals who live alone or those who are isolated due to sickness or anxiety. It could also provide inexpensive and accessible therapy services, but it can lead to addiction and further isolation. When Replika stopped allowing ‘intimate’ conversations, users lost their minds.
It is easy to say that this kind of AI companionship should not be available to children. However, neither should online pornography or social media. In the near-term, it is likely that all of us, including children, will have AI companions. For students already reluctant to navigate complex social networks, the allure of escaping to your risk-free, constant AI companion will be a far more powerful incentive than navigating who to sit with at the lunch table.
This is just one small example of the ways AI might alter how students develop and the roles schools will play. If content expertise and standardized test scores no longer hold the same value, how schools actually help students develop into responsible, grounded, and happy people could become the new premium. Many highly successful independent schools will claim they already do this and that they have nothing to lose. I would encourage these schools to take a sobering look at what their incentives, rituals, celebrations, and marketing reveals as their most important aims.
Final Thoughts
The consequences of an AI-powered world are still unknown. In a 2022 survey of AI researchers, the median results gave a 10% chance of AI wiping out mankind. Those same people also argue that AI could lead to a post-work Utopia. The reality is likely somewhere in the middle.
New York Times columnist and podcaster, Erza Klien, posed this thought experiment: What if you told someone in 1990 that they would soon have a technology that fits in their pocket and allows them instant access to all the world’s knowledge and the ability to connect over video with anyone, anywhere in real-time? Their response would likely be on par with flying cars and holodecks. Conversely, what if you told that same person that in the future there would be a pocket-sized device that increased their distractibility by at least 35% and that many would become addicted to it?
It is a sobering reminder that even the most profound advances can have complicated results. Regardless of this uncertainty, what is coming cannot be ignored. Much like a global pandemic, we cannot dismiss the possibility simply because we do not like it or because it challenges some of our fundamental understandings. As school leaders and advocates for those who learn differently, we must begin these conversations now. The traditional system has never worked for our students, and we might finally have the opportunity to redesign it.
Continue the Conversation
Continue down the AI rabbit hole as I interview some of the world’s top researchers, thought leaders and industry titans on the intersection of AI, education, and dyslexia through my online EdChat series. Guests include Debra Wilson, President of the National Association of Independent Schools, Michael Horn, Author and Co-Founder of Clayton Christensen Institute for Disruptive Innovation, Krishna Madhavan, Project Lead for AI integration at Microsoft Education, and many others.
Josh Clark is the head of school at The Landmark School, MA. Josh champions the cause of neurodiversity in education and promotes the science of reading as a vehicle for education reform and social good. He is chair of the International Dyslexia Association and an expert contributor to the global nonprofit Made By Dyslexia and Microsoft Education. Josh has presented about the importance of recognizing and supporting students with language-based learning disabilities (LBLD) all over the world. A life-long educator, he began his career in education at Lausanne Collegiate School, an International Baccalaureate World School in Memphis, TN, where he served as assistant head of the middle school and a middle and high school English teacher for seven years. Josh served for twelve years as the head of two different schools that serve students with dyslexia, the Bodine School in Memphis, TN and The Schenck School in Atlanta. Josh holds a Master of Arts in American literature from the University of the South – Sewanee and a Bachelor of Science degree in secondary education from Indiana University, Bloomington.
Josh will be the keynote speaker at the 2024 Academic Support Conference, February 11-13 in Atlanta.
