Why a person still runs the course
Teaching agricultural science at the college level is part lecture, part lab, part field. Software can write a clean summary of crop physiology or soil chemistry. It cannot stand in a greenhouse and explain why one tray of transplants is wilting while the next is fine, or check that a student can hitch an implement safely before a field lab starts. The distance between explaining a subject and supervising people learning to do the work is the heart of this job.
Two tasks make that plain. Supervising laboratory and field work means judging technique as it happens, catching unsafe habits, and changing the plan when weather, soil, or livestock refuse to cooperate. Advising students on courses, transfers, and careers means knowing a person, a program’s rules, and the employers down the road. Both are judgment calls with consequences, made in front of other people.
The research and outreach side pulls the same way. Running field trials, writing grant proposals, and working with growers and extension staff all depend on relationships and on being accountable for results. That is why this role looks different from the rest of the postsecondary teaching family only in detail, not in kind.
What AI drafts, what it assists with, and what stays with faculty
Start with the work AI can take on by itself: 2% of task time. This is the paperwork layer of teaching. Drafting course descriptions and reading lists, summarizing new literature on a pest or a fertilizer trial, and turning old notes into slides are all jobs a language model can finish to a usable standard. None of it carries the class.
Next comes the assisted share, 50% of task time, where a tool speeds a person up but the person still signs the work. Preparing lectures and labs falls here: the first draft is quick, the sequencing and the local examples are not. Grading written assignments is similar. A model can flag errors and draft feedback, while the instructor decides what a borderline paper is worth and whether the student understood the method or copied it.
The rest, 48% of task time, stays with people. Supervising field and lab work, advising students, and serving on department and curriculum committees are the clearest cases. Each one needs someone present, accountable, and able to read a room. You can see the full split in the task list above, task by task, and the method behind it on our coverage scoring page.
What the evidence says, and what it does not
Our evidence grade for this job is D. In plain terms, no study has yet tested an AI system against a qualified postsecondary agriculture instructor on this job’s actual work, so there is no parity number to report. The studies listed above sit next to this page for context, not as a verdict on classroom teaching.
What would settle it is narrow and testable: a graded comparison of AI-prepared and instructor-prepared course material judged by subject experts, measured student learning across matched sections, and some honest record of who caught problems during supervised lab and field work. Until that exists, the task mix is the better guide, and we say so rather than putting a figure on it. Our quality parity method explains why a D grade never gets a score, and the wider scoring method covers the rest.
Outside data gives some shape to the market. Employment in this occupation is about 8,920 in the United States, with median pay near $98,700 a year and projected growth of 2.9% from 2025 to 2035 (BLS, 2025). That is a small, slow-moving field. Change here shows up in hiring decisions and course loads, not in sudden exits.
When the picture could change
Most likely between 2034 and 2046 (8 in 10 of our scenarios). For how that window is built and why it is a range rather than a date, see the replacement-year method.
Two things could pull it earlier. Cost is the first: the tooling to handle this job’s software-heavy tasks runs roughly $70 to $7,430 a year, against $18,160 to $58,240 for the human hours it stands in for. The second is that almost none of this work needs a machine with hands. The physical share of the job is about 5.6%, and our robotics tier reads as none needed, so progress depends on software alone.
Two things hold it back. Accreditation and program rules still expect named faculty of record, in-person labs, and supervised field hours, and those rules change slowly. Department structure is the other brake: advising loads, committee work, and research supervision are assigned to people, and they do not get reassigned to a tool without someone taking responsibility for the outcome.
How to stay needed in agricultural education
Lean into the tasks that sit in the human column. Take the field and lab supervision nobody else wants, including the messy sites and the equipment-heavy units. Build real advising depth, so students come to you for program choices and employer leads. Say yes to curriculum work, because the person who designs the program decides what the tools are used for.
Two skills pay off. The first is applied AI literacy for your own subject: knowing where a model gets agronomy wrong, and teaching students to check it against soil tests, trial data, and extension guidance. The second is assessment design, so coursework measures what a student can actually do in a lab or a field, not what a chatbot can draft overnight.
What to do: rebuild one assignment this term so it is graded on observed practice or defended results rather than a submitted document.
Nearby roles share most of this task mix, so they are worth a look side by side: Forestry and Conservation Science Teachers, Postsecondary, Biological Science Teachers, Postsecondary, and Environmental Science Teachers, Postsecondary. You can put any two of them next to each other on our job comparison tool, see how teaching sits against other work in the education sector, or check where this kind of role lands among the jobs that mostly need a person.