How the Institute of Education Sciences is Supporting Evidence Building for Artificial Intelligence in Education

Image credit: Vanessa Loring

In May 2023, OET released our report “AI and the Future of Teaching and Learning”. This report provided several core recommendations for how to maximize the benefits and minimize the risks of AI in education settings. One of those core recommendations is to focus R&D on addressing context and enhancing trust and safety. To highlight existing programs and new funding opportunities where researchers, education leaders, and developers can come together to drive responsible innovation, we invited the National Center for Education Research, a center within the Institute of Education Sciences, to guest post about their recent funding for R&D focused on AI to improve education outcomes for students.

The conversations around artificial intelligence (AI) in education run the gamut from incredible promise for transforming education to concerns and reluctance about putting brand new and largely untested technologies in front of our nation’s youth. At the same time, product developers are moving quickly to integrate AI capabilities in their products, and some school districts and postsecondary institutions are already using large language models (LLMs) in their work. The use of AI in education is expanding rapidly, yet there are many open questions that need to be addressed, such as:

What are the uses of AI that will improve learners’ education outcomes, especially for learners who have historically been underserved by the education system?

What types of datasets are needed to support AI in education and what are the features of a high- quality dataset?

How do we ensure that AI models are developed using high- quality, representative data while also ensuring that students’ privacy is protected?

How can large language models (LLMs) be adapted for education purposes? What kinds of model tuning should be done and with what types of data sources? How should prompts be engineered?

How do we identify and mitigate sources of bias?

How do we develop AI-driven systems that prioritize trust and safety, are sufficiently transparent, and keep humans in the loop?

High- quality, rigorous research is fundamental to answering these and other open questions about the role of AI in education. The Institute of Education Sciences (IES), the research arm of the U.S. Department of Education, is the nation’s source for research, evaluation, and statistics to help educators, policymakers, and stakeholders improve outcomes for all students. Within IES, the National Center for Education Research (NCER) and the National Center for Special Education Research (NCSER) invest in high quality research and development (R&D) addressing the biggest challenges facing education in the 21st century.

Below we highlight recent IES initiatives that directly address the questions raised above, build the field’s capacity to conduct high quality R&D in this area, and provide resources for decisionmakers around the use of AI in education.

• In partnership with the National Science Foundation (NSF), IES has invested in two AI Institutes, both of which advance basic research on AI while also conducting research to understand how the new AI capabilities impact learners’ education outcomes.

1. The National Institute for Exceptional Education addresses the challenge of providing learners with timely intervention by Speech and Language Pathologists (SLPs) due to staff shortages and the time and resources it takes to conduct universal screenings and generate Individualized Education Program (IEP) plans. The Institute will develop two AI solutions: (1) the AI Screener to enable universal early screening for all children, and (2) the AI Orchestrator to work with SLPs and teachers to provide individualized interventions for children with their formal IEP.
2. The Inclusive Intelligent Technologies for Education explores how to leverage AI to support persistence, academic resilience, and collaboration within education technologies. This new generation of systems will dynamically respond to learner needs, behaviors, and development and go beyond support for domain-specific achievement. More broadly, the Institute’s work will seek to reframe how learners interact with learning technologies by prioritizing approaches that consider the whole learner.

• Through the Methods Training for Education Research program, IES invested in three data science training programs: Innovation Science for Education Analytics (ISEA): A Data Science Training Program to Advance Educational Research and Practice, Data Science for Education, and Data Science Methods for Digital Learning Platforms Training Program. Each will provide training to cohorts of education researchers interested in building their skills and capacity to use data science methods in their work.
• Through the Education Research and Development Centers program, IES has an open request for applications on Using Generative Artificial Intelligence to Augment Teaching and Learning in Classrooms (U-GAIN Center). The central purpose of the U-GAIN Center is to examine how the use of generative AI can make meaningful contributions to improving education processes and learners’ education outcomes. The U-GAIN Center will conduct research to identify the circumstances in which use of generative AI can effectively support teaching and learning, while also attending to responsible AI practices that address fairness, accountability, transparency, ethics, privacy, and security. Applications are due March 7, 2024.
• In addition to these recent initiatives, IES has been supporting R&D on AI in education since 2002 through its grant programs as well as through the Small Business Innovation Research (SBIR) program. Projects included the development and testing of intelligent tutoring systems, technology development to support automated feedback on reading and writing skills as well as math practice, and an R&D Center focused on personalizing instruction within a digital learning platform. There are also a large number of IES-funded projects studying how people learn that can be leveraged to guide the development of AI-driven education technologies.

While there is lots of promise in the AI advances for use in education, there are still many unknowns and many risks involved if we move too quickly in rolling out these technologies. High quality research, such as that supported by IES, will surface the most promising opportunities as well as help identify risks and biases in these systems and generate ways to mitigate them. To find out more about these and other IES initiatives, sign up for the IES newsflash and follow our blog.