Why this classroom stays with people
Ask whether AI will replace special education teachers in kindergarten and the task mix answers quickly: most of the day is spent in the room with five-year-olds who need physical help, redirection and reassurance. Software can draft a visual schedule. It cannot teach a child to put on a coat, or kneel beside a table and coach a tantrum back down to calm.
Two tasks make the point. Teaching self-help and daily living skills — dressing, toileting, eating, moving between activities — is hands-on, repeated and physical. So is behavior support: reading a child’s face before a meltdown, adjusting the plan mid-lesson, keeping a group safe while one child needs a minute alone. Both tasks are judgment plus touch plus presence, in that order.
The labor market around the job is steady rather than shrinking fast. The US Bureau of Labor Statistics counted about 260,870 people in this occupation with median pay of $65,120 a year, and projects employment change of -0.4% from 2025 to 2035 (BLS, 2025). That is a flat line, not a collapse, and it sits alongside long-running difficulty filling special education posts.
What AI does, what it assists with, what it leaves alone
The clearest wins are documents. Drafting and formatting individualized education program paperwork, and producing differentiated worksheets, social stories and picture schedules, are text tasks that language models handle at speed. On our read of the task list, AI can take on 8% of task time without a person in the loop. Our coverage score method explains how that share is built from task time rather than job titles.
A bigger slice is assisted work. Logging progress against IEP goals, turning a week of observations into a parent-conference summary, and translating notes for families in another language all go faster with a tool and a teacher checking it. That assisted group covers 19% of task time. The teacher still signs the document and still owns the decision.
Everything else sits with the person: 73% of task time. That group holds the instruction itself, one-on-one assessment with a child who will not sit still for a screen, supervision of a room, physical prompting, and the IEP meeting where a parent is upset and a district timeline is on the table. The task panel above shows which items land where.
What to do: let software do the first draft of the paperwork, then spend the time you save on the tasks in the needs-a-human group.
What the evidence can and cannot settle
Our evidence grade for quality against a person is D. In plain terms, nobody has tested AI head to head against a qualified kindergarten special education teacher on this job’s real tasks, so we publish no parity number for it. Treating any figure as if it were measured here would be guesswork dressed up as data.
What would settle it is specific. A blind review of AI-drafted IEPs against teacher-written ones, scored by district specialists on compliance and goal quality. A classroom trial where a tool runs progress monitoring for a term and the results are checked against teacher-collected data. Measured outcomes for assistive technology used with this age group, reported by someone other than the vendor. Until work like that exists, the honest answer is a grade, not a score. The quality parity method sets out why a D never gets a number.
When the picture could shift
Most likely between 2035 and 2056 (8 in 10 of our scenarios). The chart above plots that window, and the replacement-year method explains what it does and does not claim.
Two things could pull it earlier. Budget pressure is one: tool subscriptions cost a fraction of the staff hours they would offset, which makes paperwork automation an easy purchase for a district. Hiring difficulty is the other — when a post goes unfilled, administrators look for software to absorb the documentation load rather than the teaching.
Two things hold it back. Close to a third of this job’s work has a physical component, and the robotics panel above puts that at the dexterous humanoid tier: hardware that can safely help a small child with a coat, a spill or a bathroom trip does not exist at a school price. Accountability is the second brake. IEPs are legal documents with deadlines, signatures and dispute procedures. A district needs a named professional answerable for them, and families expect a person in the meeting.
How to stay needed
Lean into the tasks that stay in the room. Behavior intervention and de-escalation, where you read a child before the incident. Teaching self-help and social skills through repetition and physical prompting. Running the family partnership — the conferences, the home strategies, the hard conversation about a goal that is not working.
Two skills add leverage. First, assistive technology: knowing which communication device, switch or app fits a particular child, and how to set it up and train the aides around it. Second, data-informed planning: reading progress data well enough to spot when a goal needs rewriting, and to catch an AI-drafted document that says the wrong thing.
Close neighbors are worth comparing before you plan a move. Special Education Teachers, Preschool sits one step younger with an even heavier hands-on load. Special Education Teachers, Elementary School shifts toward academics and more paperwork. Kindergarten Teachers, Except Special Education covers the same age group with a larger group and a general curriculum. You can also read the wider teaching job family or the education sector page.
Next steps: put two of these jobs side by side on our compare tool, see where teaching sits among the jobs that most need a person, or read how the scoring works before you trust any of it.