House subcommittee hears witnesses on AI's promise and risks for K‑12 schools

The Subcommittee on Early Childhood, Elementary, and Secondary Education convened a hearing on artificial intelligence in K‑12 education on Oct. 12, 2025, where witnesses described both rapid technological gains and pressing policy gaps. Chairman Kiley opened the hearing by noting the speed of recent AI advances: "Artificial intelligence has been advancing at such a rapid pace ... that by the end of this hearing, anything we say this morning will probably be outdated," he said.

Witnesses described practical classroom benefits alongside systemic risks. Dr. Sid Dobrin, chair of the Department of English at the University of Florida, said the arrival of generative AI marked “a turning point in education” and argued that schools should teach students how the tools work, not merely how to use them. Dr. Julia Rafalvaire, CEO of ILO Group, recommended that the federal role be “intentionally limited” while supporting common cybersecurity and data‑privacy guidance and funding research. Erin Moe of Innovate EDU urged investments in professional development and safety frameworks. Chris Chisholm, superintendent of Pearl Public School District in Mississippi, described a district pilot that created a locally hosted AI agent to speed grading and retain veteran teachers.

The hearing framed three interlocking issues. First, witnesses emphasized instructional potential: AI tutors and agents can provide high‑dosage, individualized practice and free teachers from time‑consuming administrative tasks. The opening testimony cited research that high‑dosage tutoring can produce substantial learning gains and a McKinsey estimate that AI tools could save teachers as much as 13 hours per week.

Second, witnesses warned of equity and civil‑rights risks if federal capacity and oversight shrink. Ranking Member Suzanne Bonamici and multiple witnesses raised concerns about cuts to the Department of Education’s Office of Educational Technology and to the Institute of Education Sciences, arguing those reductions would hamper research, data availability and federal guidance that districts rely on to protect student privacy and civil rights.

Third, witnesses urged pragmatic safeguards: keeping humans in the loop for assessments and individualized education program (IEP) decisions; investing in AI literacy for educators, parents and students; and establishing state‑level vetting processes or “AI assurance labs” to evaluate tools in context. Dr. Rafalvaire and others recommended federal support focused on data privacy, cybersecurity standards and large‑scale research rather than prescriptive curricular mandates.

The hearing included district‑level examples and research references: a classroom use of Khan Academy’s Conmigo was shown as an illustration of supplemental tutoring; survey data cited in testimony reported that nearly 50% of K‑12 students use ChatGPT at least weekly and that 35% of AI‑using students frequently use it to summarize materials. Witnesses also pointed to surveys showing uneven educator adoption—about 25% of teachers report using AI in classroom preparation—and to district pilots that host models on local servers to address FERPA concerns.

No formal policy actions or votes occurred at the hearing. Committee members and witnesses repeatedly called for more research, clearer guidance on privacy and security, and sustained investment in educator training and infrastructure to avoid widening the digital divide.

The hearing record was left open for 14 days for submissions, and members were reminded they could submit statements through the clerk.