Why most of this job stays with people
Education teachers in colleges and universities train the people who will run classrooms. They teach courses on curriculum design, learning theory, and assessment. They also supervise student teaching placements, sitting in on real lessons and coaching candidates afterward. That second part is the hard part for software. Judging whether a nervous candidate handled a disruptive sixth-grade class well takes presence, context, and a professional opinion that someone will stand behind.
The advising side is similar. Faculty steer students through licensure requirements, course sequences, and career choices that depend on state rules and on knowing the student. A model can summarize a catalog. It cannot vouch for a candidate to a district hiring manager, or decide that someone is not ready to lead a classroom yet.
Plenty of the surrounding work is text, though, and text is where current systems are strongest. Syllabi, reading lists, rubrics, slide decks, and grade records all sit in that category. That is why the share of task time our method counts as automatable is not small. You can see the split above; the method behind it is explained in how coverage is measured.
What AI does, what it helps with, what it leaves alone
Routine document and record work is the automatable slice. Drafting a syllabus from a course description, building rubrics, assembling reading lists, and keeping attendance and grade records are all tasks a competent assistant tool can produce a usable first version of. That group accounts for 2% of task time in our task split.
A larger set of tasks moves faster with help but still needs the instructor’s judgment. Preparing lectures and seminar materials, summarizing new research in education, and writing first-pass comments on student papers all fit here. The instructor edits, corrects, and decides. That assisted group covers 51% of the work.
Then there is the part that stays with a person: observing and coaching candidates during student teaching, advising students on licensure paths, and serving on curriculum and accreditation committees where a named faculty member signs off. Those tasks make up 47% of task time, and they are the reason this page reads the way it does.
What the evidence actually shows
The evidence grade for this occupation is D, which is our lowest. That means no study has tested an AI system against qualified teacher educators on this job’s own tasks, so we publish no parity number for it. Claims that models match or beat professors here are not supported by a measured comparison.
What would settle it is specific. A blind trial where faculty and a model both write feedback on the same student teaching observations, scored by trained reviewers. A comparison of candidate pass rates on state licensure exams between AI-supported and faculty-led sections. Published accreditation findings on programs that shifted supervision to software. Until something like that exists, the honest answer is that the quality question is untested. Our general approach is set out in the scoring methodology, and the parity rules in is it better than a person.
When the picture could change
Most likely between 2034 and 2045 (8 in 10 of our scenarios). What the window measures is explained on the replacement year method page.
Two things could pull it earlier. Cost is one: the AI tooling for this job’s automatable tasks runs roughly $80 to $7,800 a year, against $14,820 to $47,570 for the human hours it touches. No hardware is needed either, since none of the work is physical in the robotics sense, so there is no machine to buy or maintain. Online and hybrid teacher-prep programs are the other accelerant, because more of the teaching already happens through text and video.
Two things hold it back. Accreditation and state licensure rules still require supervised clinical practice signed off by a qualified person. And the job is not shrinking much: the BLS counts 60,830 of these positions in the United States with median pay of $75,350 (BLS, 2025), and projects employment up about 2.5% from 2025 to 2035 (BLS projections). Slow institutional procurement and faculty governance add further drag.
Good to know: the pressure here shows up first in course load and section size, not in whole positions disappearing.
How to stay needed in teacher education
Lean into the tasks that sit in the human group. Take on clinical supervision and classroom observation rather than handing it off. Own student advising, especially the licensure and placement conversations. Sit on the curriculum and accreditation committees where program decisions are made and defended.
Two skills pay off. First, assessment design: writing tasks and rubrics that measure what a candidate can actually do in front of students, which also makes AI-written submissions easier to spot. Second, practical fluency with the tools, so you can set program policy on them instead of reacting to it. Our guide to AI skills employers want covers what that looks like in practice.
If you are weighing a move, nearby roles are worth a look: Psychology Teachers, Postsecondary, Career/Technical Education Teachers, Postsecondary, and Instructional Coordinators. You can see how this role sits against its peers on the postsecondary teachers family page or across the education sector. To put two jobs side by side, use the job comparison tool, or browse the jobs that mostly need a person list.