Why the pattern still comes off a bench
Ask will AI replace patternmakers and the honest answer starts with the object itself. A pattern is a physical master. Someone lays out the dimensions, machines the sections, then fits and hand-finishes the halves until a casting or molded part pulls free without tearing. Software can calculate that geometry. It cannot feel a draft angle that is a hair too tight.
Material does not behave the way a drawing says it will. Metal shrinks, plastic warps, and a mounted pattern that worked last year wears at the parting line. Much of the trade is correcting for that in the moment: shaving a fillet, easing a surface, deciding whether a worn pattern is worth repairing or should be cut again. Repair work is the hardest part to script, because every damaged pattern fails in its own way.
The shop economics matter too. This is a small occupation, with about 1,470 US jobs (BLS, 2025) and median pay near $58,000 a year. Most work is one-off or short-run. The robotics read on this page puts most of the job in physical space, and the machines that do that work are fixed automation: CNC mills, routers and printers that a person sets up, fixtures, tools and all, for each new job. Setup time is where the hours go.
What AI drafts, what it checks, and what people still do
Start with the tasks AI can take outright. These are the paper-and-screen steps: building the model from a drawing, computing shrink and machining allowances, and turning the finished geometry into toolpaths. Share of task time in that group: 0%. On our coverage scale, which runs 0 to 100 and is explained on the coverage method page, this job sits at 14.
Then there is the assist work. Software is useful for comparing measured dimensions against spec, flagging a feature that will not draw cleanly, suggesting an operation sequence, and pricing out stock and material. A patternmaker still signs off, because the model does not know what the last three jobs on that machine taught you. Share of task time in that group: 29%.
What is left is the bench and the machine. Fixturing and setting up cuts, fitting pattern sections together, finishing surfaces by hand, mounting patterns on plates, and repairing wear all sit with a person. Share of task time in that group: 71%. That split is why the headline figure lands where it does: 80 out of 100 (higher is safer).
What the evidence does and does not show
No one has tested AI against a qualified patternmaker on real pattern jobs. That is why the evidence grade here is D, and why this page carries no parity number. A grade like that means not measured, not measured-and-close.
A real test would not be hard to design. Give a model and a journeyman the same set of jobs from a working shop: a new mounted pattern, a short-run plastic master, and two repairs. Then count what the shop counts, which is first-article pass rates, rework hours, dimensional checks against print, and how many patterns the foundry accepts without a callback. Until something like that is published, we hold the line rather than guess. You can read how we grade parity on the quality-parity page, and the wider method at how our scoring works.
When patternmaking could change, and what moves the date
Most likely after 2046 (8 in 10 of our scenarios). The replacement-year method page explains what that window is measuring and how wide it is on purpose.
Two things could pull it closer. Cheaper additive printing of sand molds and cores removes the need for a hard pattern on some jobs altogether, which cuts the task rather than automating it. And tighter CAD-to-CNC pipelines keep shifting hours from the bench to the screen, where software is already competent.
Two things hold it back. Fixed automation still needs a person per job, so a shop running dozens of short-run parts never reaches the volume that justifies full automation. And the capital is thin: small job shops buy a machine when a job pays for it, not on a software release cycle.
There is a separate pressure worth naming. BLS projects employment in this occupation falling 22.8% between 2025 and 2035. Fewer jobs is a different story from automated jobs. Demand shifts, offshoring and parts that no longer need a pattern can shrink a trade while the remaining work still needs skilled hands. The task-level read and the headcount read can point in different directions at the same time.
How to stay needed in the trade
Lean into the parts of the day that nobody has automated. Setup and fixturing, because that skill transfers across every machine in the shop. Fitting and finishing, because tolerance by feel is still scarce. And repair, because a shop that can save a worn pattern on short notice keeps the foundry running.
Two skills pay for themselves. First, CAM and CNC programming, so you own the screen work instead of handing it to someone else. Second, metrology: CMM, scanning and first-article inspection, which turns you into the person who proves a part is right.
What to do: Ask your shop which jobs now start as a scan or a printed mold, and get your name on the first one.
Close trades are worth a look if you want options. Compare the task mix with Model Makers, Metal and Plastic, Patternmakers, Wood and Tool and Die Makers. The wider picture sits on the metal and plastic workers family page and in manufacturing. To see where hands-on trades land overall, read jobs that mostly need a person, or put two of these side by side on our compare tool.