Why the teaching hour stays with a person
Self-enrichment teachers run classes people choose for interest rather than credit: guitar, pottery, yoga, beginner coding, conversational Spanish, cooking. Will AI replace self-enrichment teachers in those rooms? The task mix points somewhere narrower. The teaching hour itself stays with a person, while much of the prep around it moves to software. Students attend by choice and can quit any week, so holding a group together is part of the work, not a soft extra.
Two tasks on the list above show why. The first is demonstrating a technique and then watching the student attempt it: a wrist angle, a knife grip, a note that drifts flat. That loop needs eyes, ears and hands in the same space. The second is adapting instruction to each student’s pace and ability. A class of eight beginners contains eight different problems, and the instructor reads them in real time from posture, hesitation and the look on a face.
There is also the part nobody writes down. People sign up for a hobby class to be around other people. The instructor sets the tone, settles the nervous beginner and keeps the confident one from taking over. Software can deliver a curriculum. It does not hold a room.
What AI does, what it helps with, and what stays with people
Start with the share AI can take on its own: 13%. This is the desk side of the job. Drafting a lesson outline for a six-week beginners’ course, writing handouts and practice sheets, producing class descriptions, building schedules and keeping attendance and progress records are all text and data work that current tools handle quickly.
Then the assisted share: 21%. Here the instructor stays in charge and the tool speeds things up. Preparing teaching materials for different skill levels is faster with a generator. Monitoring student progress is easier when an app logs practice time or checks pitch and tempo between lessons. The instructor still decides what the data means for next Tuesday.
The rest is the part that needs a human: 66%. Demonstrating skills in person, correcting form on the spot, managing group dynamics and keeping people enrolled sit here. That is also why this job’s score on can AI do it today, 27 out of 100, stays well below the paperwork-heavy jobs. The coverage method explains how that share of task time is estimated.
What has actually been tested
Not much, in this specific job. The evidence grade for is it better than a person is D, and a D means the comparison has not been measured, so we publish no parity number for self-enrichment teaching. Plenty of studies look at AI tutoring for school subjects with gradable answers. Hobby instruction is different: the outcome is a skill performed, plus whether the student comes back.
A clean test would compare an AI-led beginners’ course with an instructor-led one over the same weeks, measuring skill gain judged by independent assessors, completion rate and re-enrollment. Until something like that exists, treat confident claims in either direction with care. Our scoring method is published so you can see exactly where the gaps are, and the what the AIs say list shows how the assistants themselves answer.
What would move the date
Most likely between 2035 and 2052 (8 in 10 of our scenarios). The replacement-year method sets out how that window is built.
Two things could pull it earlier. Cost is one: the panel above compares software licensing against paid instruction hours, and community centers and gyms watch that gap closely. Group size is the other. If a provider accepts an app plus one floor supervisor for twenty students instead of two instructors for ten, the headcount falls without any breakthrough in teaching quality.
Two things hold it back. Roughly a quarter of these tasks have a physical component, and the robotics panel places the hardware needed at the dexterous humanoid tier, which is not a near-term purchase for a yoga studio or a pottery co-op. Demand is the second brake. People pay for a class partly to be taught by someone who is good at the thing. Remove the instructor and you have changed the product, not just the staffing.
The wider market is steady rather than shrinking. About 332,110 people held this job in the US, median pay was $46,800, and projected employment change for 2025 to 2035 is 3.5% (BLS, 2025). Modest growth, with the usual caveat that fewer new openings hit newcomers first.
How to stay needed
Lean into the tasks that sit in the human column. Teach in person and make the demonstration the center of the session. Give correction while the student is holding the instrument, the brush or the pose. Run the group as a group, so people enroll again because of who they sat next to.
Two skills are worth building. The first is using AI tools for prep, so your planning and materials take an hour instead of four and your paid time goes to teaching. The second is simple business craft: pricing, scheduling, retention and getting referrals. Independent instructors who can fill their own classes are far less exposed than those waiting for a center to assign them hours.
What to do: put one AI tool to work on your lesson outlines and handouts this month, and spend the saved hours on live coaching or filling the next course.
Nearby jobs share a lot of this task mix. Compare the scores for Tutors, Adult Basic Education, Adult Secondary Education, and English as a Second Language Instructors and Substitute Teachers, Short-Term, or look at the whole other teachers and instructors family and the education sector page for the pattern across teaching work. You can also put two jobs side by side on the compare tool, or see where in-person instruction lands on the list of jobs that mostly need a person.