EXPOSED
Lecture slides, problem sets, solution keys, syllabi, grant boilerplate and literature reviews are already produced at usable quality by current models, and much of an engineering professor's week is exactly that. What resists is running a physical lab, supervising senior design and thesis work, judging which research directions are worth funding, and being the named mentor whose recommendation letter and advising a student's career depends on. Scored for the modal tenure-track/teaching-faculty worker; adjunct lecturers teaching high-enrollment intro courses sit meaningfully lower, since online AI-assisted course delivery attacks that tier first.
Mixed — a routine tier and a judgment tier. Statics problem sets, thermo lecture decks, ABET syllabus mapping and NSF broader-impacts boilerplate are the recurring deliverables of a semester and a model drafts them at usable quality, which caps this at 11; what holds it above the automatable band is standing in the lab when a student's tensile rig is misconfigured, scoping a senior design project that must actually build, and calibrating how hard to push a struggling MS candidate.
Some physical or field component. You are in the machine shop, the wind tunnel, the clean room or the concrete lab several hours a week — signing off on instrument use, spotting the unsafe fixture, running the demo that doesn't work on the projector — but grading, proposal writing, committee service and lecture prep are screen work, which is why this lands at 10 rather than the 15+ of a field engineer.
No licence, no signature requirement. No licence gates the classroom: most engineering faculty never hold or maintain a PE, ABET accredits the program rather than certifying the individual instructor, and nothing you sign carries personal legal exposure — the 3 reflects only the institutional gatekeeping of the PhD and tenure review, not any statutory protection of the work itself.
The human relationship is the product. The recommendation letter with your name on it, the lab you invited a student to join, and the advising conversation that redirects someone from grad school to industry are not substitutable outputs — students choose research groups by advisor, not by curriculum — but this sits at 15 rather than 19 because the 200-seat sophomore lecture is a delivery channel where the relationship is thin and your identity is close to interchangeable.
Meaningful discretion. You decide which research direction to spend three years and a graduate stipend on, whether a thesis is defensible, whether an academic-integrity case goes to the dean, and how a senior design team's failing prototype is graded — real ambiguous calls with career consequences for others, though bounded by department curricula, promotion committees and IRB/EHS review rather than borne alone, which is what keeps it at 13.
Has AI actually changed your work?