Three Strategies to Enhance Accessibility Using AI
Artificial Intelligence (AI) has the potential to close achievement gaps between teaching and learning, but only if it is implemented correctly and with accessibility embedded from initial development to ensure access to each user. Ignoring the need to address learner variability introduces a risk of widening the achievement gap for students with disabilities, culturally and linguistically diverse students, and other populations who have historically been underserved.
In exploring AI and accessibility, CoSN connected with Bruce Alter, a content expert who serves as an Assistive Technology consultant with the Tigard-Tualatin School District and the Woodburn School District in Oregon. Bruce has experience informing Oregon school leaders from these districts to state officials regarding the potential of AI to enhance learning for students with disabilities, and he suggests a path for school districts to start implementing AI in a way that promotes accessibility and educational equity.
“GenAI offers potential for students with disabilities that they never had before. Using these tools, they could communicate, create art, music, and write beyond anything they could do previously. But, for this to occur, school districts need to have policies on appropriate use of GenAI and, most importantly, all educators need ongoing professional development on how to use these tools to improve their student outcomes”.
The implementation and to AI tools in schools is crucial and urgent. Some students and teachers have already started exploring the possibilities. However, with a lack of structured guidance, there can be a missed opportunity to advance and potentialize the skills of all students and, in the long term, widen the gap of inequity. For instance, speech recognition systems in virtual assistants have become an essential tool to create captions or translations in real time. However, when individuals with speech disabilities like deaf accent or dysarthria provide input, the performance of these systems fall short. (Morris, 2020). Recognizing this gap, Bruce has started to test a speech recognition software designed for non-standard speech with one of his students. By doing so, his student can now communicate, engage, and interact with her learning environment much better.
Nonetheless, there is still a long journey ahead. At this pivotal moment, we must “level the playing field” and iterate Bruce’s efforts with all students. This can be achieved by opening access to AI and creating possibilities for students with disabilities to go beyond what their current assistive technology allows them to do today. Bruce provides a plan of action to make this shift happen and segments it into short-, medium-, and long-term strategies towards implementation in schools.
Short-Term Action Toward AI Implementation in Schools
“We must provide our teachers with actionable, professional development resources that should include understanding GenAI limits and biases.”
As a first step, professional development is key. Education leaders should understand what generative AI is, how it works, and how to implement it responsibly with students. Even when GenAI systems provide answers to almost everything, they are far from perfect. Most of these tools perform, learn, and generate responses informed by datasets, which are provided by the developers of each application. Because the datasets tend to lack diversity in their user population samples, some AI tools might not work correctly for some students with disabilities. For instance, there are tools that can identify objects for people who are blind and describe them. Still, as the tool’s algorithm was trained with images captured by sighted users, it lacks efficacy when applied with pictures taken by blind users, which tend to have lower quality (Morris, 2020). Moreover, there are other major risks associated with the potential re-identification of students or tools having a negative bias towards individuals with disabilities. Considering the latter, professional development is not only necessary to understand the responsible use of GenAI within the classroom, but also forges a path to standardize the potential and limits of implementation in the classrooms all over the district.
An Initial Resource for AI
At this primary stage, Bruce recommended seeking support from organizations that are already connected with experts in the AI field and that can provide professional development tools. Specifically, he references AI for Education, a nonprofit that focuses on providing AI literacy training at the school and district-level as well as AI policy development tools, and AI adoption strategies. On the organization’s website, it is possible to find introductory workshops such as “Intro to GenAI for Educators” and “Using AI for differentiated instruction.”
Medium-Term Action Toward AI Implementation in Schools
After creating a general awareness of AI within educators, Bruce suggests three actions to take in the medium term. Firstly, it is essential to seek guidance from teachers and researchers knowledgeable about learning processes that enable students to fully develop their skills. Through the experts’ support, education stakeholders can establish when students can use these tools to augment rather than replace their abilities. Bruce exemplifies, “we should specify writing skills they need to have developed before they can use writing support tools like spell check, word prediction, grammar check, and GenAI tools”. Secondly, it is essential that special education students are guaranteed access to AI tools that enhance their skills, even when general education students are not allowed to use them. As a means towards accessibility, educators must also have access and be encouraged to implement GenAI tools in their classes. In the context of Individualized Education Programs, teachers should be motivated to use AI tools and envision them in the classroom as an effective form of assistive technology. For example, using GenAI to modify assessments, readings, and curriculum according to the students’ Lexile level. Thirdly, it is necessary to scale the professional development to not only special education teachers but staff, general education teachers, paraprofessionals and all the individuals in the district who are linked to the students’ learning process in some form.
Long-term Action Toward AI Implementation in Schools
“We must consider our education has to change in light of these tools being an accepted and ubiquitous part of modern society.”
Moving Toward Universal Access of AI Tools
As Artificial Intelligence tools are becoming embedded in our society, it is necessary to create a culture around the implementation of AI tools in education and provide reasonable guard rails through policy adoption and professional development. Education stakeholders should envision how all students, teachers, and staff can have access to AI tools in their daily study and practice. For this to happen, districts and educators can rely on guidelines such as while assessing the implementation of these tools while ensuring that, by using AI, educators are creating learning environments where barriers are reduced, and all students have the potential to succeed.
References:
Morris, R. M. (2020). AI and Accessibility: A discussion of ethical considerations. Communications of The ACM, 63(6), 35-37.
Author: Fernanda Pérez Perez, 2024 Blaschke Scholarship Fellow
