The University of California, Irvine School of Education has received a $10 million, five-year grant from the U.S. Department of Education's Institute of Education Sciences to establish a national research center on generative AI in postsecondary writing instruction. The WRITE AI Center will combine research, technology development, and educator training, with a focus on building evidence rather than simply encouraging AI adoption.
The center's plans include a national study of the AI tools colleges are already using, further development of a guided writing platform called PapyrusAI, and a randomized controlled trial across six community colleges starting in year three. UC Irvine will lead a partnership involving MDRC, the American Council on Education, the California Community Colleges' Digital Center for Innovation, Transformation and Equity, the Virginia Community College System, and Houston City College.
Tamara Tate, Associate Director of UC Irvine's Digital Learning Lab, wrote on LinkedIn that the funding was "the sole award made through IES's competition and its national research and development center focused on generative AI in postsecondary education."
A research agenda built on open questions
The center's first task is a deceptively basic one: finding out what generative AI tools colleges are actually using. WRITE AI plans an ongoing national study covering the tools in use, where and how they are deployed, instructional features, costs and infrastructure requirements, staff training needs, approaches to safety and fairness, and the evidence currently available about their impact.
Those findings will feed into a publicly accessible, searchable resource for researchers and practitioners. The research then moves from mapping current practice to testing an intervention. During years one and two, researchers will conduct exploratory work and iterative development. Year three is scheduled to include a randomized controlled trial in first-year composition courses across six community colleges.
The trial will examine writing quality and AI literacy alongside student and instructor self-efficacy, access to personalized writing instruction, course retention, student persistence, and later achievement in writing-intensive courses. Many of the center's proposed benefits remain research questions rather than established outcomes. WRITE is explicitly setting out to test whether AI-supported approaches affect learning and how institutions can implement them responsibly.
Mark Warschauer, Distinguished Professor at UC Irvine and Director of the Digital Learning Lab, described the center's evidence-building role on LinkedIn. "Through research, tool development, partnerships, and practical resources, the WRITE AI Center will help postsecondary institutions make evidence-based decisions about AI while facilitating effective, responsible, and equitable writing approaches," he wrote.
PapyrusAI keeps instructors in control
A central part of the work involves PapyrusAI, a web-based generative AI environment originally developed by a UC Irvine research team with National Science Foundation funding. Rather than giving students unrestricted access to a general-purpose chatbot, the platform uses embedded, research-based prompts designed to structure AI asking questions instead of providing answers. Students can use it while planning, drafting, revising, and reflecting, while instructors determine which tools are available and adapt them to course objectives and student needs.
The prompts are designed to encourage students to explain decisions and reflect on their work rather than simply receiving completed writing. AI literacy is built into the same structure - students learn to prompt, question, and evaluate AI responses, including recognizing output that is incomplete or incorrect.
PapyrusAI also includes a multi-agent system for scoring and providing feedback on student writing in alignment with instructor expectations. The center is studying how instructor-facing analytics could provide information about student needs and AI use, rather than presenting that functionality as a finished, evidence-backed intervention.
For institutions, the platform currently supports Google sign-in and access through Canvas. Planned development includes interface improvements, expanded prompt libraries, improved instructor dashboards, additional learning analytics, stronger pedagogical guardrails, and integration with Canvas SpeedGrader. The center describes instructor judgment as central to the model. Its research agenda includes examining how AI can provide personalized support without replacing the writing process itself, alongside questions about data protection, fairness in AI-generated feedback, transparency, and students' ability to question AI output.
Training and collaboration alongside the research
WRITE AI is also building a training and capacity-development program for researchers, educators, and institutional leaders who need to evaluate generative AI in education. Its work will include professional learning, instructional resources, and guidance derived from the research program. A UCI professional-learning course on AI in Higher Education is scheduled to begin on September 29, 2026, for higher education instructors and instructional-support professionals. For educators building these skills now, an AI Learning Path for Teachers covers practical classroom applications.
The partnership brings together expertise in writing research, postsecondary education, educational technology, research and evaluation, and institutional leadership. Writing advisors named in the project include Anna Mills, Skip MacArthur, and Steve Graham. The center's first two years will be dedicated to exploratory research and development of PapyrusAI before the planned six-community-college randomized trial begins in year three. Year four will focus on case studies and continued implementation, followed in year five by final analysis, dissemination, and implementation guidance. The center says its findings will ultimately inform an Implementation Support Guide for institutions considering AI-supported writing tools.
Why this matters for educators
For instructors and instructional designers, the center's work offers something the field has lacked: controlled data on whether AI-supported writing instruction actually improves outcomes. The randomized trial across six community colleges will produce findings on writing quality, AI literacy, retention, and persistence that institutions can use when deciding whether and how to adopt these tools. The center's commitment to instructor control in PapyrusAI also signals a design direction worth watching - AI configured as a structured assistant that students question and evaluate, not a black box that produces finished work. For those making decisions about AI tools in their own classrooms, the AI for Education resources track these developments as they unfold.
Your membership also unlocks:
