Why this teaching work stays with people
Will AI replace postsecondary earth science teachers? The honest answer is that parts of the job are already shared with software, while the parts that define it are not. These faculty prepare and deliver lectures in atmospheric science, geology, oceanography and astronomy, then supervise students through laboratory and field work. A model can draft a lecture outline in seconds. It cannot stand on a shoreline transect, watch a student mishandle a core sample, and correct the technique before the data is ruined.
The second anchor is advising. Faculty counsel students on course choices, graduate programs and research directions, and they sit on committees that judge whether a thesis is sound. That work depends on knowing a particular student over several years, and on being accountable to a department for the judgment. Accreditation, grading appeals and research supervision all put a named person on the record.
The job is also small and relatively well paid, which shapes how much pressure it feels. The Bureau of Labor Statistics counted about 9,900 of these postsecondary teachers in the United States, with median pay of $103,170 and projected employment growth of 2.4% from 2025 to 2035 (BLS, 2025). Slow growth and tight hiring budgets matter more here than any single tool. Our full method is set out at how we score jobs.
What AI does, helps with, and leaves to people
Some administrative work now runs with little human input. Compiling reading lists and bibliographies, and keeping attendance and grade records, are the clearest examples: both are text and data handling with a fixed format. Across this job, the share of task time where AI can take the lead sits at 8% of the work we score.
A larger block is assisted rather than handed over. Preparing lecture notes and course materials, and marking written assignments and exams, are faster with a model in the loop, but a faculty member still sets the reading, the rubric and the final grade. Keeping up with new research in atmospheric or marine science is the same story: search and summary tools speed the reading, the judgment about what belongs in a course does not transfer. The assisted share is 54%.
Then there is the part that stays with a person: supervising undergraduate and graduate lab and field work, and mentoring student research toward publication. That share is 38%. Our Can AI do it? figure for the whole job is 39 out of 100, and the method behind it is explained on the coverage scoring page.
What has actually been tested
Our evidence grade for Is it better than a person? is D. A grade of D means there is no direct, published test of an AI system against qualified faculty in this occupation, so we publish no parity number for it. That is a gap in the record, not a verdict either way.
What would settle it is specific: a graded comparison of AI-written and faculty-written course materials judged blind by subject experts; a controlled study of AI marking against faculty marking on the same earth science assignments, with agreement rates reported; and measured learning outcomes for students supervised through field or lab sequences with and without AI support. Until work like that exists for this field, the quality question stays open. The standard we apply is on the quality parity page.
When the picture could shift
Most likely between 2034 and 2046 (8 in 10 of our scenarios). What the window measures, and how we build it, is described on the replacement year method page.
Two things could pull the change earlier. Online and hybrid delivery keeps growing, and automated course content plus automated assessment fit that format well. Budget pressure on small departments is the second: where a program is already thin, administrators look for ways to teach the same material with fewer faculty hours.
Two things hold it back. The physical share of the work is small, so robotics is not the barrier here, and the barrier is instead institutional. Accreditation rules, credit-hour requirements and faculty governance change slowly, and a human instructor of record is usually required. Research supervision is the other brake: grant-funded labs and thesis committees are built around named advisors with professional accountability. Compare this job with a neighbor on the side-by-side comparison tool.
How to stay needed
Lean into the tasks that carry your name. Run the field and lab sequences, and own the safety, instrument and sampling training that goes with them. Take on thesis and dissertation supervision rather than only lecture hours. Keep advising load, because departments protect people who retain students.
Two skills pay off. The first is working data fluency in your subfield, from climate model output to satellite and sensor records, so you teach the pipeline students will actually use. The second is assessment design: writing tasks that test reasoning in the field and the lab, where a generated answer is easy to spot.
What to do: rewrite one course this year so at least one assessment depends on data the student collected or interpreted in person.
If you are weighing nearby options, the closest work sits with Environmental Science Teachers, Postsecondary, Geography Teachers, Postsecondary and Physics Teachers, Postsecondary. For the wider picture, see the postsecondary teachers family, the education sector page, or what chatbots themselves say about these roles in our what the AIs say list. The Still needs a human figure for this job is 65 out of 100 (higher is safer), and every job is searchable in the full job rankings.