This arXiv paper examines how U.S. state education departments have written guidance on AI use in K-12 schools. More than half of states have issued formal guidelines since ChatGPT's public release in late 2022, and the authors treat these largely independent documents as a record of how states responded to a shared technological disruption. They analyze publicly available guidance from 35 states and Puerto Rico.
The authors compare the documents along several dimensions:
- pedagogy and instruction
- audience
- actionability
- workforce and career readiness
- responsible use and equity
- alignment with state demographics and administrative capacity
They also look at who wrote each document and which sources it cites, aiming to identify the epistemic foundations behind its claims.
Priorities converge on topics anchored in law, with data privacy and academic integrity each addressed in 97% of documents. Most documents mention algorithmic bias (97%) and equity of access (86%), though fewer discuss specific marginalized populations. The authors report a common framing of urgency around AI integration, tied to workforce readiness (86%) and to a loosely defined idea of personalized learning (86%). They find this urgency isn't backed by specific guidance on capacity-building and accountability, and the documents offer little in the way of actionable direction.
The authors found no relationship between state priorities and state demographics or capacity, and none with partisanship. What they did find were interdependencies between states. Many share citations to nonprofits such as TeachAI and Code.org, and some reproduce TeachAI's model guidance nearly word for word. Citations to other states concentrate on a few early publishers, such as Oregon and Washington. No state lists parents as contributors to its guidance, and only 5 include students.