Matthew Marino, a professor at the University of Central Florida who studies the intersection of technology and special education, is excited about the potential of generative artificial intelligence for teaching and learning.
AI is “going to open a whole new front for education,” he said.
The sentiment is shared by a majority of parents and educators of students with learning disabilities, according to a report from the Special Olympics Global Center for Inclusion in Education. The survey shows that parents and educators believe AI will make learning more accessible, inclusive, and personalized. But they’re also concerned about AI’s potential to decrease human interaction and leave some schools and students behind.
Marino underscores that there are ways AI shouldn’t be used in instruction, especially because there still isn’t enough data on how using generative AI for instruction could affect students with disabilities. Most datasets that AI tools are trained on contain a lot of information about neurotypical students, but they don’t have nearly as much to draw from on students in special education, which could translate into technology that isn’t as helpful for those populations.
Marino spoke with Education Week about why he’s excited about AI for special education, what research is out there, and how educators can use the emerging technology.
This interview has been edited for brevity and clarity.
Could you explain why you think AI is ‘going to open a whole new front for education?’
The capabilities we now have with AI are absolutely amazing. Students no longer need to struggle to write a paragraph, for example. All they have to do is create a prompt, and a large language model can write for them. I’m not saying it’s a tool that’s going to be used in place of learning to write, but it’s a tool enabling students who have disabilities to spend more time refining their thinking than on generating basic text content. I have a student with muscular dystrophy who can’t physically write a paragraph. It allows students like him to spend time evaluating their thoughts and organizing them in a way where they can be communicated effectively.
Why is that approach more beneficial?
A lot of times, kids will get stuck with the first task, where they have to take information that they’ve learned in one context and transfer the knowledge into a paper. This type of assignment is common in classrooms across the country. When students have to generate text from scratch, they often shut down and do nothing or they become a behavior problem for the teacher.
What we’re trying to do is give them a venue where they can generate text information and reflect on it. We want them to develop a thought process following the scientific method. They’re using the problem-solving method to analyze text that’s been generated by AI.
What research is already out there on using generative AI for special education students?
There’s very little published peer-reviewed research on AI and education. The reason is because of the rapid development cycles associated with AI. There’s new software coming out every week, with major revisions occurring about every six months. Right now, most of the studies are exploratory case studies. There are no experimental or quasi-experimental studies because of the nature of research projects and how long they take.
By the time researchers get through the [institutional review board] approval to actually do their research, and carry out the study, the AI software is already dated. We submit the results for peer review. It’s in review for six months, and then by the time it’s actually published, the software is really outdated. That’s a big challenge that we’re going to have to work through as we move forward. The old school way we’re doing research, carrying out research projects, and publishing the information has to change in order for us to stay current with what we’re actually doing research on.
It shouldn’t be seen as a replacement for effective teachers or reason for students to not work hard. We should be using AI to help students with things like executive-function skills.
Some teachers are hesitant to use AI in their work or with their students. What would you say to them?
I would say I’m extremely excited about it. The capabilities of AI are absolutely amazing.
I’ll give you an example. I was working with a student not long ago, and we were looking for some graphic organizers to help the student to understand the differences between a plant cell and an animal cell. After we spent some time on the internet, I started thinking, “Well, maybe I should use AI and see if AI can create it for me.” So I just typed in a quick prompt and said, “Create an infographic comparing animal and plant cell organelles.” Within 15 seconds, I had a beautiful graphic with everything I wanted for the student. It would have taken an additional 10 or 15 minutes to find a good graphic without the AI. The amount of time teachers can save is amazing, not only on the student instruction side but also on the assessment side.
Do you have an example for the assessment side?
I can give you an example from a special education administrator’s perspective. We did a study last winter looking at a school district in the northeast [United States]. The school district was using Excel spreadsheets to collect IEP services data from case managers in the district. The district administrators were using the data to develop budget and staffing projections for the school board.
In a district where you have 100 case managers, each with 20 students on their caseload, you have a pretty big data set. Administrators can spend months synthesizing and analyzing the data. The administrators in this study were able to use a generative pretrained transformer, or GPT, called “data analyst.” By just typing in a prompt, data analyst was able to take all of the different Excel spreadsheets and combine the data into a compelling narrative with graphics to support staffing decisions. It’s normally a three-month project. They did it using data analyst in three days.
Are there ways that AI should not be used with special education students?
It shouldn’t be seen as a replacement for effective teachers or reason for students to not work hard. We should be using AI to help students with things like executive-function skills. That means planning and organizing their day, coming up with tasks, using the reminders features, and using text-to-speech. It’s not going to replace the thought process students go through. It’s a tool they will have on their phones and in their pockets for life.
Do you have any other suggestions for special education teachers looking to try AI?
Teachers should look strategically at their students and [ask themselves]: If I spend my time using artificial intelligence with the student, what are the benefits and limitations going to be and what is the return on investment?
Once they’ve made the decision to use AI, they should develop a plan where they implement the AI in a consistent way over time. Then they can evaluate if it improved the student’s performance. It has a lot of potential.
