Why the glass stays in human hands
Will AI replace glaziers? Not in the way headlines suggest. Almost all of a glazier’s day is physical work on a site that never repeats: measuring an opening, cutting glass or a substitute to fit, lifting it into a sash or frame, and sealing it so the building stays dry. Software can read a spec sheet. It cannot hold a 200-pound lite steady on a ladder while a sealant bead sets.
The share of task time that still needs a person here is 91% of the job. That reflects the mix of work: removing broken glass and old putty, fastening panes into frames with clips, molding or sealant, and fixing leaks where a seal has failed. Each of those tasks depends on what the technician finds when the old unit comes out. Frames are out of square. Fixings are corroded. Measurements taken from a drawing do not match the wall.
Smart glass is a real change, but it changes the product, not who installs it. Switchable and sensor-linked glazing still arrives as a heavy, fragile panel that someone has to carry, set, wire where needed, and weatherproof. If anything, more electronics in a unit means more care on site and more diagnosis when something stops working.
What software does, what it assists, and what it leaves to the crew
The tasks our review puts fully within reach of AI today account for 0% of task time, and they sit at the desk end of the job: turning a sketch or spec into a cut list, and working out the glass type, thickness and quantity an order needs. That is document work, and language models handle document work well. You can see how we draw that line on our coverage method page.
A larger slice of the AI-reachable work is assistance rather than replacement, at 9% of task time. Tools help with measuring and marking an opening when a laser measure feeds straight into a takeoff, and with selecting a unit for a given opening and climate. The human still signs off, because a wrong number on a custom lite is a wasted panel and a return visit.
What is left is the job itself. Cutting and grinding glass to size, setting it into storefront, curtain wall or residential frames, and sealing the joint stay with people. So does removing and replacing damaged glass after a break-in or a storm, often in an occupied building with customers on the other side of the opening. None of that is a text problem.
What has actually been tested
Evidence for how well machines perform glazing work against a qualified person grades D. In plain terms: there is no direct, published test of AI or robotics doing this trade next to a working glazier, so this page gives no parity number for it. The benchmarks that exist measure writing, coding and office-document tasks, not setting a pane in a frame.
What would settle it is specific. A timed trial of a machine system against a two-person crew on real openings, measured on fit, seal integrity, breakage rate, callbacks and cost per unit installed. Until something like that is published, the honest answer is that the physical side is untested. Our quality parity method explains why we refuse to put a figure on an untested claim, and the wider approach is set out in our scoring methodology.
Outside our scoring, the labor market data is worth knowing. The Bureau of Labor Statistics counted 58,480 glaziers employed in the United States, with median pay of $57,080 a year, and projects employment growth of 1.9% between 2025 and 2035 (BLS, 2025). That is slow growth, not decline.
When glazing work could change
Most likely after 2048 (8 in 10 of our scenarios). The reasoning behind that window is on our replacement year method page.
Two things could pull that forward. The first is prefabrication: most of the automation in glass today is fixed plant in factories, where cutting, coating and sealing units already run on machines. The more glazing that arrives as a finished, pre-glazed panel, the less cutting and fitting happens on site. The second is cheaper material handling, such as mobile vacuum lifts and semi-automated placement rigs that cut the crew size on big curtain wall jobs.
Two things hold it back. Site variation is the big one: no two retrofits, storefronts or historic frames present the same problem, and the physical share of this job’s tasks is high. The other is simple arithmetic. A capable mobile robot has to beat a skilled crew on cost, breakage and speed across hundreds of awkward openings, and the equipment bill for that is far heavier than the software bill for a cut list.
How glaziers stay needed
Lean into the tasks that need judgment on site. Complex retrofits and restoration work, where the opening is never square and the old unit has to come out cleanly. Structural and storefront installation, where sealing and weatherproofing decide whether the job holds. And diagnosis: finding why a unit is fogging, leaking or binding, which is a reading of the whole assembly, not one part.
Two skills raise your floor. First, digital measurement and estimating: laser tools, takeoff software and model files, plus the habit of checking what the file claims against the wall. Second, crew leadership and customer handling, because the person who can quote, schedule and explain a repair becomes hard to route around.
What to do: ask for the storefront, curtain wall and restoration jobs, not just the straight residential swaps.
Nearby trades sit in similar territory if you want to compare the work. See Carpenters, Roofers and Sheet Metal Workers. You can put any two jobs side by side on our job comparison tool, read the wider picture for this trade group on the construction trades workers family page, or see how the whole construction sector looks. For context on hands-on work generally, our list of the jobs that mostly need a person is a useful next stop.