Helping in-service teachers bring generative AI into classrooms safely — the ExLAIM program (NC State) (NSF AI grant $1M)

The NSF awarded about $1,000,000 to NC State's "Beginnings: Experiential Learning for In-Service Teachers: Augmenting Teaching and Learning with Generative AI" (NSF Award 2526340; program: IUSE / ExLENT; January 2026 – December 2028).

1Preparing teachers for generative AI

Per the abstract, the project addresses the critical gap in K-12 educator preparation for integrating rapidly evolving generative AI (GenAI) tools into practice. While GenAI promises to transform learning, most secondary teachers lack both the technical proficiency and pedagogical frameworks to use these technologies effectively and safely.

The project develops a four-week "experiential learning in AI and MerryQuery (ExLAIM)" program bringing together in-service K-12 teachers, undergraduate computing majors, and industry mentors to co-design AI-integrated curricula, assessments, and implementation plans tailored to classroom contexts.

2A learn-use-reflect cycle

Building on project-based and active learning, participants follow a learn-apply-reflect cycle, using MerryQuery, a retrieval-augmented-generation chatbot, to scaffold lesson planning and assessment design. Groups iteratively prototype AI-enhanced instructional modules that integrate safety considerations, outlier detection, and reliability checks for classroom implementation.

Structured mentorship with faculty, graduate students, and industry mentors reinforces skill acquisition. Resources are shared nationally as open access through the Computer Science Teachers Association (CSTA), expanding AI professional development to under-resourced and geographically dispersed districts.

3Learn, use, reflect — and repeat

This four-week experiential programme is not built to hand over knowledge in lectures but to run three stages repeatedly.

1. 1LearnUnderstand how generative AI works and where it fails
2. 2UseDraft lesson plans and assessments with a retrieval-augmented chatbot
3. 3ReflectRevisit safety considerations, outlier detection and reliability checks against the classroom

Serving K-12 teachers, undergraduate computing students and industry mentors in one room, the programme has them co-design a curriculum fitted to the context of a classroom. That the people who build and the people who use sit together, rather than a teacher learning alone, is the point of the design.