Why the bench holds the job together
Jewelry work is small, physical and unforgiving. A jeweler sizes a ring to a customer’s finger, cuts the shank, aligns it, solders the joint and polishes the seam until it disappears. Stone setting is the same story at a smaller scale: raising beads, bending prongs, seating a stone that cost more than the tool holding it. Software can model all of that. It cannot do it on a worn, thin, hundred-year-old band sitting on the bench in front of you.
The second thing protecting this trade is the object itself. Most repair work arrives as someone else’s property, often with sentimental value and no replacement. The jeweler inspects it, decides what the metal will take, and tells the customer what is worth doing. That judgment runs on touch and experience, not on a prompt.
That does not mean nothing shifts. Design, quoting, cataloging and customer messaging are already moving toward software, and mass-market casting has leaned on CAD and 3D printing for years. The honest picture is erosion at the edges of the role rather than the bench disappearing.
What software runs, what it assists, what people keep
Start with the share of task time that software can run on its own: 4%. That slice is office work in disguise. Generating design variations from a reference image, writing item descriptions for a listing, pricing a job from metal weight and labor, and keeping records of repairs and inventory all sit here. None of it touches a torch.
The assisted slice is bigger: 13%. CAD modeling for cast pieces is the clearest example, with the jeweler judging wall thickness, prong height and wearability before anything is printed. Gem examination is another. Imaging and measurement tools can flag inclusions, proportions and likely treatments, but a grader still signs the call and a bench jeweler still decides how the stone will be held.
The rest, 83% of task time, is the work people keep. Soldering and laser welding on damaged mountings. Setting stones by hand. Finishing, polishing and final inspection. Talking a customer through what their grandmother’s ring can and cannot survive. The task list above shows how those land.
Good to know: the biggest near-term pressure on this trade is not a robot at the bench; it is fewer shops and fewer apprentice slots.
How strong the evidence is
No one has run a clean head-to-head test of AI against a working jeweler. Our evidence grade for “Is it better than a person?” is D, and a D grade means not measured, so no parity number is published for this job. We say that plainly rather than filling the gap with a guess. You can read how the grades are set on the quality parity method page.
What would settle it? Three things. A published trial comparing machine-graded and human-graded stones on the same parcels, with disagreement rates. A measured comparison of robot and human stone setting on real mountings, including scrap and breakage. And a study of CAD-plus-generative design against trained designers, judged on pieces that are actually wearable, not renderings. Until work like that exists, the score for this occupation leans on task structure and the physical demands of the role, which is set out in the scoring method.
The market context is separate from the AI question. The Bureau of Labor Statistics counts about 22,440 US jobs in this occupation, with median pay around $52,540 and projected employment falling roughly 3.2% between 2025 and 2035 (BLS, 2025). Shop closures, imports and retail consolidation drive most of that, not automation at the bench.
What would move the date
Most likely after 2046 (8 in 10 of our scenarios). The chart above shows the spread, and the replacement year method explains what the window measures.
Two things could pull it earlier. First, real progress in dexterous robot hands: the robotics panel on this page puts most of the work in the physical column at the dexterous humanoid tier, which is exactly the capability labs are chasing. Second, more production moving to cast-and-print pipelines, where a designer and a machine replace several bench hours per piece.
Two things push it back. Repair work is non-standard by nature, and every damaged mounting is a one-off, so there is no clean training set and no repeatable fixture. And the economics are awkward: the cost panel above compares tooling against a trained jeweler’s wages, and a two-person shop doing resizes and prong rebuilds has little reason to buy a robot that handles only the easy cases. Trust matters too. Customers hand over irreplaceable property and expect a person to answer for it.
How to stay needed in this trade
Lean into the work that stays human. Repair and restoration on damaged or antique pieces. Hand stone setting, especially pave, channel and bezel work on difficult mountings. Final finishing and inspection, where the difference between good and sellable is visible in seconds.
Two skills are worth adding. The first is CAD and print-to-cast workflow, so you own the design-to-bench handoff instead of losing it to someone who does. The second is gemstone identification and appraisal, including lab-grown detection, which keeps you in the loop on valuation as imaging tools spread.
If you are weighing adjacent paths, the closest work sits nearby: gem and diamond workers, etchers and engravers, and grinding and polishing workers, hand. You can also see where this job sits among other production occupations and across manufacturing.
Next step: put this job beside one of those on the job comparison tool, or browse the jobs that mostly need a person list to see which hands-on trades sit near it.