Inside IES Research

Notes from NCER & NCSER

AI-Augmented Learning for Individuals with Disabilities: New Funding Opportunity, Current Research, and the Potential for Improving Student Outcomes

This March, IES Director Mark Schneider released a blog in which he discussed exploring a partnership with the National Science Foundation (NSF) to encourage scientists with expertise in AI and related fields to address the important post-pandemic need for accelerating learning. IES is now excited to announce our resulting participation in NSF’s National Artificial Intelligence (AI) Research Institutes—Accelerating Research, Transforming Society, and Growing the American Workforce solicitation. In this blog, we describe this new funding opportunity, provide examples of existing NCSER-funded research in this area, and highlight the potential for such research to further improve outcomes for learners with disabilities.

 Artificial Intelligence Research Funding Opportunity

With funding from the American Rescue Plan, NCSER plans to support research under Theme 6, Track B: AI-Augmented Learning for Individuals with Disabilities. Proposals must discuss how the work will respond to the needs of learners with or at risk for a disability in an area where the COVID-19 pandemic has further widened existing gaps and/or resulted in decreased access and opportunities for students with disabilities to learn and receive support services. Please review the solicitation, the webinar (November 16), and the frequently asked questions for more information. Interested applicants should note the primary focus of this institute:

The primary focus of an institute in AI-Augmented Learning includes research and development of AI-driven innovations to radically improve human learning and education. Achievement and opportunity gaps, particularly for learners from disadvantaged or underserved communities, have always been present, but COVID-19 has exacerbated them. Institute plans for this theme should address and measure outcomes with direct education impact, in both the short- and long- term, that have practical significance to educators, parents, or other decision-makers. Plans must also directly address algorithmic bias, model transparency, security and data privacy in the support of learning.”

Current NCSER-Funded Grants Applying Artificial Intelligence and Machine Learning

Prior to the new collaboration between IES and NSF, NCSER funded several grants that apply artificial intelligence and machine learning approaches, including those described below.

With a 2018 NCSER grant, Dr. Maithilee Kunda and her team at Vanderbilt University are building on a technology-based intervention known as Betty’s Brain. This computer-based instructional program for middle school science, designed with the support of a 2006 NCER grant, allows students to teach a computer agent to understand certain concepts, increasing their own knowledge and understanding. Dr. Kunda and her team are developing a new game called Film Detective, which is designed to improve theory of mind (ToM) reasoning in adolescents with autism spectrum disorder (ASD). More information about this project can be found in this IES blog.

With a 2021 NCSER grant, Dr. Patrick Kennedy and his team at University of Oregon are using machine learning to validate a well-known assessment, Dynamic Indicators of Basic Early Literacy Skills, 8th Edition® (DIBELS 8) as a screener for dyslexia. As of 2020, 47 states require that students be screened for dyslexia in early elementary school and many states use DIBELS for this screening. However, it remains to be validated for this purpose. To address the validity of the DIBELS for screening, this research team is using machine learning approaches to predict and classify scores in relation to a pre-defined target. This will allow the research team to draw conclusions about the validity of the DIBELS 8 for dyslexia screening. These conclusions will be disseminated widely to state and local education agencies and other stakeholders.

The Potential of AI for Improving Outcomes for Learners with Disabilities

In addition to the work that IES is funding, AI has already demonstrated potential for improving outcomes for learners with disabilities in many other ways:

  • AI has been used to support children with ASD who have difficulties understanding people’s emotions, with AI-driven apps and robots helping students practice emotion recognition and other social skills.
  • AI has informed the development of algorithms that can help those involved in assessment identify disabilities in students, such as ASD, specific learning disabilities (dyslexia, dysgraphia, and dyscalculia), and attention-deficit/hyperactivity disorder (ADHD).
  • AI-embedded interventions have included error analysis to inform instruction and personalized feedback in spelling and math for students with disabilities.

Despite these advancements, there appear to be persistent gaps in AI research for students with disabilities, such as AI for students with intellectual and developmental disabilities. This is an especially important area of work because many of these learners have multiple disabilities and/or serious health conditions. For example, children with intellectual and developmental disabilities who also have hearing loss or visual impairment have compounded challenges. Some students with Down syndrome also have hearing loss and other health complications, such as cardiac issues. AI affords an opportunity to integrate health information across different applications to improve the quality of life for these students. These technological solutions can assist in managing information about the students and communicating health information between teachers, physicians, and caregivers.

AI has the potential to transform special education. We hope that this NCSER-NSF partnership will encourage researchers to be creative in planning projects that move the field of AI forward as well as provide innovative solutions to support learners with disabilities.

This blog was co-authored by Sarah Brasiel (Sarah.Brasiel@ed.gov), program officer at NCSER and Bennett Lunn (Bennett.lunn@ed.gov), Truman-Albright Fellow for NCSER and the National Center for Education Research (NCER). IES encourages special education researchers to partners with experts in Artificial Intelligence to submit to this NSF AI Institute solicitation 22-502 to increase the evidence base on use of AI for this population.

Film Detective: How an AI-powered Game Aims to Improve Outcomes for Students with ASD

Artificial intelligence (AI) is poised to revolutionize the way humans live, even in ways yet unseen, and education is no exception. IES funds research at the cutting edge of technology and education science, and, as Director Mark Schneider has recently pointed out, AI may eventually serve to help educators identify, assess, and support students with disabilities. In 2018, NCSER awarded funding to Dr. Maithilee Kunda of Vanderbilt University to do just that.

Dr. Kunda and her team are developing a new game called Film Detective to improve theory of mind (ToM) reasoning in adolescents with autism spectrum disorder (ASD). ToM reasoning is the ability to infer the mental state of others, allowing us to understand and predict behavior based on our perception of their beliefs, intentions, and desires. The game builds on a technology-based intervention known as Betty’s Brain. Developed with support from a NCER grant, Betty’s Brain is a computer-based instructional program for middle school science that allows students to teach a computer agent to understand certain concepts. By teaching the agent, students grew their own knowledge and understanding. Dr. Kunda and her team are building on this software by adapting the learning-by-teaching model to improve ToM reasoning in neurodiverse students. (For more on Dr. Kunda’s perspective on the importance of neurodiversity, see this blog.)

The Film Detective game takes students through an interactive storyline in which they must help a scientist from the year 3021 “decode” the way people in today’s world behave in a series of films. The stakes are high as students help a scientist unlock a time machine by retrieving codes hidden in films by an evil scientist—aptly named Von Klepto—who has stolen items from the Museum of Human History. By teaching the computer agent—the player’s robot sidekick (named T.O.M.)—how to identify modern behaviors, the student develops their own ToM reasoning. The Film Detective storyline is a product of the creative talents of several Vanderbilt creative writing students, and the game mechanics were designed with insights of college students with ASD themselves. With the help of post-doctoral student and project lead, Roxanne Rashedi, the project team has used participatory design and qualitative methods to better tailor the game to the community for which it is intended. By working closely with students with ASD and their families, the project team was able to refine the original Betty’s Brain structure with new reward structures and storylines that balance the challenge of the game with the frustration that students can feel playing the game.

Screen shot of the Film Detective’s theatre and time machine room
Film Detective’s Theatre and Time Machine Room (illustration by Kayla Stark)

Every part of the project draws on the diverse expertise of the team, and the inclusion of a variety of perspectives has been crucial to informing the project’s development. The team includes experts from Vanderbilt’s School of Engineering and the Vanderbilt Medical Center’s Treatment and Research Institute for Autism Spectrum Disorders (TRIAD), with Dr. Kunda and students in computer science and psychology providing insights in cognitive science and artificial intelligence. The joining of expertise in artificial intelligence, clinical psychology, and educational psychology has allowed the team to merge theoretical perspectives on ToM development with conceptions of knowledge representation and modeling in computational systems. This approach offers the team a unique framework for understanding the development of social reasoning skills in students with ASD. Beyond the theoretical, the team has also leveraged artificial intelligence to evaluate how students progress through the game, using advanced data mining techniques and eye-tracking-enabled user studies to better understand how students with ASD can develop greater ToM reasoning through learning-by-teaching.

Film Detective’s hallway to concessions
Film Detective’s Hallway to Concessions (illustration by Kayla Stark)

The work that has gone into Film Detective exemplifies the ways that novel research that combines technological advancement and diverse perspectives can lead to important innovations in the education sciences. While Film Detective is still under development (it is currently being user tested, and readers are encouraged to sign up to take part here), IES is eager to see what will come out of this exciting collaboration.

Dr. Maithilee Kunda is the director of the Laboratory for Artificial Intelligence and Visual Analogical Systems and a faculty investigator for the Frist Center for Autism and Innovation at Vanderbilt University. This blog was written and edited by Bennett Lunn, Truman-Albright Fellow for the National Center for Education Research and the National Center for Special Education Research.