■ HIGH RISK ■ Construction & Trades
The cutting shed is automating; the wall isn't. CNC saws and robotic carvers now do the shaping, but setting stone on an actual building — plumb, weathertight, and matched to hundred-year-old work — remains stubbornly human.
“CNC stone cutting is precise. Your chisel is charming but slow.”
Our AI replacement risk score — how we score jobs
Stone masonry splits into two worlds that outsiders lump together. In the shop, masons cut, shape, and carve stone — sills, copings, ashlar blocks, ornamental work — traditionally with saws, chisels, and a lifetime of feel for how a particular sandstone splits. On site, they lay and set that stone: building walls, cladding facades, repointing historic joints with lime mortar, and restoring cathedrals and courthouses where the new work has to sit invisibly beside the old.
The shop side is where automation has already moved in. Five-axis CNC saws and robotic milling arms cut profiles, mouldings, and even figurative carving from digital models with tolerances no hand can match, and large fabricators run them around the clock. A capital, a baluster, or a run of matching window surrounds that once took a banker mason weeks now comes off a machine in days, needing only hand-finishing. Digital scanning of deteriorated ornament, fed straight to the mill, is standard practice in big restoration fabrication. That's a real contraction in shop hours, and it's why our score sits at 62 rather than down with the plumbers.
Site work resists for reasons robots hate: every wall is different, access is by scaffold, stones weigh what they weigh, and the judgment calls — which stones to replace versus retain, how to match a mortar that's been weathering since 1890, how to bed an irregular block so it drains — are made in place, by eye and hand. Heritage restoration adds a layer of conservation ethics and craft knowledge that's genuinely scarce; many countries report shortages of qualified masons for their backlog of historic buildings. The trade's future is smaller on the production end and secure on the restoration end.
Automatability: our editorial assessment of current and near-term AI capability
Shop fabrication is consolidating around CNC now — that shift is well underway and keeps compressing bench-mason hours through 2030. Site and restoration work faces no comparable threat this decade; if anything, demand outruns the supply of qualified masons. Expect the trade to bifurcate by around 2030: machine-assisted fabrication with fewer hands, and a protected, shortage-driven craft tier on the scaffold.
The production-carving side is shrinking — machines genuinely cut and carve faster and more accurately from digital models. But the trade as a whole isn't dying: setting stone on buildings and restoring historic masonry are shortage occupations in much of the world. It's less a dying trade than one migrating from the bench to the scaffold.
Yes — robotic mills carve capitals, tracery, and even statuary from 3D scans, and large restoration projects use them routinely. What they produce still typically needs hand-finishing for texture and undercuts, and the decisions about what to carve, in which stone, to match which weathered original, remain human. The robot is a very fast apprentice, not a master.
Yes, with the right emphasis. Train in setting, repointing, and conservation work alongside cutting, and learn the digital fabrication tools rather than resenting them. Heritage bodies in the UK, Europe, and North America consistently flag mason shortages for restoration backlogs. The apprentice who combines craft hands with CNC literacy has the strongest position in the trade.
Anything done on a building rather than in a shop: laying, setting, repointing, structural repair, and conservation assessment. Those tasks happen in unstructured environments with one-off judgment calls — the conditions robots handle worst. Bench production of repeatable profiles is the exposed end; bespoke restoration on scaffolding is the defended end.