Welder

When it comes to automation, welding is really two jobs. Most welders work in factories: BLS counts 63% of the 437,700 welding jobs in manufacturing. Factories that make the same part over and over can hand much of that welding to machines and robots, and BLS says "automation in manufacturing may limit overall demand" for welders. That is automation of real, core work, and it is why this is a 2 and not a 1 like electricians or plumbers. The other kind of welding happens in the field and in repair shops: pipelines, bridges, building frames, ships, one-off repairs. BLS describes welders working outdoors in all weather, high up on scaffolds and in awkward positions. That kind of work is very hard to automate. A 2025 McKinsey study of robots in construction lists welding among the jobs general-purpose robots might one day help with. But it says large-scale use of humanoid robots on building sites "may be a decade away," because they still struggle on busy, messy worksites. It expects them to support workers rather than replace them. AI chatbots barely touch this work: a 2025 Microsoft Research study of how much of each job's tasks chatbots can help with put welders in the bottom quarter of 785 jobs.

BLS projects only 2% growth from 2025 to 2035, slower than the average for all jobs. But it still expects about 40,300 openings a year, mostly to replace welders who retire or change careers. It also expects rebuilding aging bridges, highways and buildings to keep demand up. That fits the rubric's "2 Low": machines take some factory work, but the core of field and repair welding stays human.

What would change this score. If portable robot welders that can be set up quickly on a job site become common in construction and shipyards, or factory welding jobs shrink fast enough to crowd new welders out, this should go to 3. For a new welder, structural, pipe and repair work are the safer path than production-line welding.