Why engineering faculty work stays with people
Readers ask whether AI will replace engineering teachers because so much of the job arrives as text. Lecture notes, slides, problem sets, rubrics, written feedback, grant drafts: language models handle that kind of material well, and fast. That is the honest pressure on this job, and it is pressure on tasks rather than on the whole role.
The rest of the work is harder to hand over. Someone has to be in the lab while a team wires a test rig or loads a beam, and decide on the spot whether what students are about to do is safe. Someone has to supervise capstone and thesis projects, where the answer is not in any textbook and the judgment call is about method, not arithmetic. Someone has to advise students on course loads, internships and whether a design idea is worth another semester. Those tasks are conversations and responsibility, not output.
Market conditions matter too. Federal data puts US employment in this occupation at about 40,270, with median pay of $109,270, and projects 7.8% growth from 2025 to 2035 (BLS, 2025). Growth that steady does not look like a job being emptied out. It looks like a job where the paperwork and prep get faster, and where departments think harder about how many instructors they need for the same number of sections.
What AI does, what it assists, what it leaves alone
Start with the tasks AI can already carry on its own: drafting course outlines, generating practice problems and worked examples, assembling reading lists, and turning last year’s notes into this year’s slides. Those tasks make up 2% of task time. Coverage, our answer to whether AI can do the work today, reads 37 out of 100 for this occupation, and you can see how that figure is built on the coverage method page.
Next come the tasks where AI assists and a person signs off. First-pass grading of problem sets, feedback on student writing, literature searches for a research proposal, and keeping lecture material current with new standards all fall here. The instructor still sets the question, checks the marking and owns the grade. Those tasks account for 60% of task time.
Then the work that stays with the instructor: lab and studio supervision, capstone and graduate project advising, academic and career counseling, and the departmental and accreditation work that decides what a degree means. That group is 38% of task time. It is also where most of the job’s value sits for students, which is why the headline score lands at 66 out of 100 (higher is safer).
How strong is the evidence?
Weak, and we say so. The evidence grade for this occupation is D, which means no study has tested AI against qualified postsecondary engineering instructors on the real job. There are plenty of studies on AI tutoring and on AI writing code, but teaching a junior-level mechanics course, running a lab section and supervising a capstone team are not the same tasks, so we do not carry over a result that was measured somewhere else.
Because of that, we publish no parity number here. What would settle it is specific: a controlled comparison of AI-generated course material against instructor-designed material on student learning outcomes in the same program; blind marking trials on real engineering coursework, including partial credit and design work; and measured outcomes for AI-supervised versus faculty-supervised lab and project work. Until something like that exists, the grade stays where it is. The quality parity method page explains how grades move, and the full methodology shows the rest.
When the picture could shift
Most likely between 2034 and 2045 (8 in 10 of our scenarios). What that range measures, and how it is built, is set out on the replacement year method page.
Two things could pull it earlier. Cost is the first: the tool spend we track for the automatable slice of this job runs from about $80 to $7,780 a year, against $23,020 to $78,680 for human time on the same tasks. Second, none of that slice needs robot hardware. It is desk work, so the usual physical brake does not apply here, and adoption can move at software speed.
Two things hold it back. Labs, studios and project reviews need a responsible adult in the room, and safety and liability sit with a named person. And curriculum change runs through accreditation reviews, faculty governance and tenure structures, which move in years, not quarters. You can set this job beside a neighboring one on the compare tool to see how different those brakes are.
How to stay needed in an engineering department
Lean into the tasks that stay. Take the lab and studio sections seriously, including the safety judgment that comes with them. Supervise capstone and graduate projects, where the work is open-ended and the student needs a person who will argue with them. Keep the advising load, because that is the relationship students remember and the one departments cannot buy.
Two skills are worth real time. One is assessment design that tests reasoning rather than output: oral defenses, in-lab checks, design reviews, problems with messy or incomplete data. The other is practical fluency with AI tools, enough to teach students how engineers use them and where they fail, which is now part of the subject itself.
What to do: rewrite one course’s assessment so a student cannot pass it by submitting generated text, and keep the change in your teaching file.
Nearby jobs face the same split in different proportions: Computer Science Teachers, Postsecondary, Physics Teachers, Postsecondary and Architecture Teachers, Postsecondary. For the wider picture, see the postsecondary teachers family, the education sector, and our list of jobs that mostly need a person.