When teacher Mary Acebu starts her workday in Bay Point, Calif., she describes a schedule that often runs long—arriving early to prepare for students and taking paperwork home late. Over the past two years, she has turned to artificial intelligence as a way to move faster through tasks that can otherwise consume evenings, including work tied to students’ individualized education programs, or IEPs. In interviews with NPR, Acebu said the time savings changed what she could do with that extra availability: she could spend more time interacting with students in class instead of working through documents on a computer.
Acebu’s experience reflects a broader staffing and workload strain in special education, where districts nationwide have struggled with hiring and retaining teachers. In the 2024-25 school year, NPR reported that 45 states said they faced special education teacher shortages, with conditions often worse in schools serving low-income students. Educators say some of the burden comes from paperwork that piles onto teaching responsibilities, including legally required documentation that outlines goals and services for students who qualify for special education.
NPR reported that special education teachers increasingly use AI to help with IEP development—work that educators and families treat as detailed planning for students’ academic and functional progress. Acebu and others describe IEP goals as central to translating learning plans into classroom support. NPR also described the role of researchers Olivia Coleman of the University of Central Florida and Danielle Waterfield of the University of Virginia in studying AI’s potential in special education planning and implementation.
Acebu said she takes a cautious approach, using AI alongside a requirement for what she calls a “human touch.” NPR reported that she began by taking courses on safe and effective AI use and that her district, Mt. Diablo Unified, entered agreements with companies offering education-focused AI tools, including MagicSchool AI and Google. In her account, she customized chatbots for her school using district-vetted processes and trained them on state standards, assessments and other special education data so the tools could support her work rather than replace her judgment.
In NPR’s reporting, Acebu described using her “little assistants” for tasks ranging from creating personalized worksheets to developing IEP goals, while also double-checking outputs and incorporating her own professional decision-making. Coleman and Waterfield, whose research NPR summarized, said teachers use AI for multiple components of the IEP workflow, including writing goals, tracking progress, synthesizing data and creating differentiated learning materials. Acebu, who NPR reported has earned a doctorate in instructional technology and serves on an AI task force developing a district policy, said she still treats the final step as a human responsibility.
Not all teachers in Acebu’s school started as early adopters. NPR reported that Paul Stone, a special educator at Riverview for 22 years, said an increased number of students he serves created major stress and raised questions about whether his job could be split between paperwork and direct work with children. After a tutorial from Acebu, Stone said he tried the chatbot and found it helpful as a time saver. He told NPR he still reviews information before using it, and he said he has used AI to produce simple summaries of complicated data for parents at IEP meetings.
Even as educators describe potential time relief, advocates warn that AI introduces risks that can matter as much as efficiency. Ariana Aboulafia, NPR reported as lead author of a Center for Democracy and Technology report, characterized AI tools as “a Band-Aid” for special education teachers who feel overworked and said guardrails are necessary. NPR said Acebu also highlighted student privacy as a top concern, warning teachers not to put information that could identify students into tools that are not appropriately protected.
Coleman, Waterfield, CDT and other researchers described to NPR that AI use varies widely across classrooms, including instances where teachers use free consumer platforms alongside district-approved tools. NPR reported that Aboulafia said AI models can be biased, including against people with disabilities, and raised concerns about the inherent mismatch between AI’s pattern-based operation and the individualized process that special education law requires. Aboulafia said CDT’s research found some teachers using AI entirely to develop IEPs, and she said there must always be a “human in the loop.”
In Acebu’s classroom, NPR reported, the intent of IEPs is to connect written plans to daily instructional support. She described tracking a student in her eighth-grade class, saying that after a period of intensive support tied to IEP goals, he is reading now. Acebu said she now begins class about 30 minutes before students arrive and leaves after the last bell, and she described the change in work-life balance and teaching time as part of why AI has mattered in her day-to-day role.