Why Trades Workers Are Safe from AI in Every Country We Measured

What makes trades work AI-resistant

There are four structural reasons trades work resists AI displacement. They apply consistently across every country in the WorldJobsData dataset - from electricians in Germany to plumbers in Vietnam to construction workers in Nigeria. The underlying conditions are the same because the work itself has the same characteristics regardless of geography.

Physical-spatial complexity

No two construction sites are identical. A plumber sent to install a new bathroom in a 1960s apartment block will find pipes that are not where the original plans say, walls that have been modified without documentation, and access routes that require improvisation. An electrician rewiring a Victorian house encounters a different set of undocumented modifications, hidden junctions, and non-standard fittings in every room. The spatial reasoning required - understanding a three-dimensional space in real time, identifying where constraints are, deciding how to route a cable or pipe through that specific environment - is something current AI cannot perform. AI can assist with planning, generate estimates, or help with documentation. It cannot stand in a roof space and work out where to run the new circuit.

Tool adaptation

A tradesperson's skill is not just knowing what to do. It is knowing how to do it with whatever tools and materials are on the site, in the specific conditions of that site. A carpenter fitting a door frame in a building that has settled over decades uses different techniques than in a new build. An HVAC engineer diagnosing a heating fault adapts their approach based on the sounds, temperatures, and visible symptoms they encounter. This kind of adaptive tool use - selecting from a repertoire, improvising when the standard approach doesn't fit the situation - is not something AI can replicate in physical environments. The knowledge is embodied in hands and eyes and experience, not in a textbook or a training dataset.

Error stakes

A wrong cut in electrical wiring has immediate physical consequences. A poorly installed gas fitting creates a hazard that may not manifest for months. A structural error in a load-bearing element can cause collapse. The stakes of errors in trades work create a high bar for any replacement technology. In low-stakes environments - content generation, data classification, document drafting - AI errors are correctable at low cost. In trades work, errors have irreversible physical consequences. This risk profile means that even where AI could theoretically assist with specific sub-tasks, the liability and safety implications of deploying it in high-stakes physical environments have slowed adoption severely.

Embodied knowledge

Trades expertise is not in textbooks or training manuals. A master electrician's knowledge of how to interpret the condition of a circuit by the feel of a connection, or a plumber's understanding of water pressure by the sound of flow through a pipe, cannot be extracted into a dataset and replicated by an AI system. This embodied knowledge - acquired through years of physical practice in varied real-world conditions - is precisely the component of trades work that AI has no pathway to replicate with current architectures. AI can generate plumbing installation guides. It cannot plumb a bathroom.

The AI score vs robotics score distinction

The 2.5/10 AI score for ISCO 7 trades workers and the 4.5/10 robotics score sit alongside each other in the WorldJobsData dataset, and the gap between them matters. AI exposure measures the risk from digital and cognitive automation - software systems that can replicate information processing, communication, and routine cognitive tasks. Robotics exposure measures the risk from physical automation - machines that can replicate physical manipulation tasks in structured environments.

The higher robotics score (4.5/10) reflects the fact that some specific sub-tasks within trades work have been successfully automated in controlled environments. Rebar tying robots exist and are deployed on some large construction sites. The Hadrian X bricklaying robot, developed by FBR in Australia, can lay bricks at scale on straightforward external walls under controlled conditions. The SAM100 (Semi-Automated Mason) from Construction Robotics in the US can increase bricklaying productivity on certain commercial projects. These are real deployments with real productivity gains.

But they are narrow, specific, and not generalisable. Hadrian X cannot enter a building, navigate a staircase, or work around the complications of a real construction site interior. SAM100 works on specific geometries of commercial wall construction - not on residential brickwork, curved walls, or sites with access constraints. The variability that defines most real-world trades work - the reason each job site is different - is precisely what makes broad robotics deployment in construction impractical for the next decade or more.

The 2.0/10 AI score for building trades specifically (ISCO 71) is the more reliable signal for most trades workers. The robotics risk is real but distant for the broad population; the AI risk is low and remains so for the foreseeable future.

The shortage and safety combination

The most important fact about trades work is not just that it is protected from AI - it is that it is simultaneously in structural shortage across the economies where WorldJobsData has the most detailed data.

In the US, Associated Builders and Contractors (2026) estimates the construction industry needs approximately 500,000 additional workers immediately, on top of normal hiring to replace attrition. This figure does not include the shortfall in specialty trades like electrical and HVAC, where the shortage is proportionally even larger relative to the qualified workforce available. The US has over 143.1 million workers total. The construction sector's inability to fill roles is not a demand problem - construction activity is at a historically high level. It is a supply problem caused by a generation of workers choosing other career paths and a pipeline of apprenticeships that has not kept pace.

In Germany, the Bundesagentur fur Arbeit (2024) reports 86,000 unfilled apprenticeship positions in skilled trades - a figure that understates the actual shortage because it counts only positions formally registered, not the informal gap between what employers need and what they can find. Germany's construction and renovation sector faces acute shortages of electricians, plumbers, heating engineers, and roofing specialists. The shortage is structural because German trades training requires multi-year apprenticeships, and the pipeline cannot be expanded quickly even if more young people chose to enter.

In the UK, the Construction Industry Training Board (CITB, 2024) projects that the construction sector needs 225,000 additional workers by 2027 to meet planned infrastructure and housing programmes. The UK government's target of 1.5 million new homes by 2029 depends on a trades workforce that does not currently exist at the required scale. Electricians, plumbers, bricklayers, and general construction workers are all in shortage across most UK regions.

This combination - low AI displacement risk AND rising structural demand - is not common in the labour market. Most occupations face at least one of these pressures: either AI exposure or market saturation. Trades occupations face neither. The result is a labour market position that is unusually strong from both sides simultaneously.

The earnings story

Trades work does not just offer structural protection - it offers above-median earnings in every economy where WorldJobsData has wage data, and those wages are rising as shortages intensify.

In the US, the Bureau of Labor Statistics OEWS (2024) reports a median annual wage for electricians of approximately $62,600. Plumbers, pipefitters, and steamfitters have a median of approximately $61,550. HVAC mechanics and installers have a median of approximately $57,300. All three figures are above the US median annual wage of approximately $48,060 (BLS, 2024). Experienced electricians and master plumbers in high-demand metropolitan areas frequently earn above $80,000.

In Germany, OECD Average Annual Wages data (2024, USD PPP) puts skilled trades earnings at approximately $46,000-$52,000 for experienced workers - above the German median in purchasing power terms and rising faster than inflation as shortages intensify. In the UK, ONS ASHE (2024) data shows construction and building trades workers earning median annual wages of approximately $42,000-$48,000 GBP equivalent, again above the UK median.

The pattern is consistent across economies: trades offer above-median income with below-median AI risk. In a labour market where above-median wages are increasingly associated with high-exposure professional roles (managers, ICT professionals, finance workers all score above 5.5/10 on AI exposure), trades represent a pathway to comparable earnings with fundamentally different risk exposure.

Which trades are most and least protected

Sources: WorldJobsData ISCO-08 scoring model. AI scores based on task-content analysis. Robotics scores based on physical automation feasibility assessments.

The pattern within trades is consistent: AI exposure is low across all sub-groups (2.0-2.5/10). Robotics exposure varies more, reaching 6.0/10 for food processing trades where factory-scale automation is already established. But even the highest robotics score in this group (6.0/10) is lower than the AI exposure scores for managers (5.5/10) and ICT professionals (8.5/10). Trades workers face less displacement risk from both AI and robotics than most professional and clerical occupations face from AI alone.

The building and electrical trades (ISCO 71 and 74) have the lowest scores on both dimensions. These are also the trades with the most acute shortages globally. The combination of lowest displacement risk and highest unmet demand makes building and electrical trades the most structurally durable positions in the WorldJobsData dataset.

What this means for trades workers and career changers

If you are already in trades: the data confirms what experienced tradespeople tend to already know intuitively. Your job is not going to be automated away in the next decade. The more useful question is how to position within the trades for the work that will be in highest demand - new-build electrical work for EV charging infrastructure, heat pump installation for the energy transition, building retrofit work for decarbonisation. These are trades sub-specialisms where demand will grow significantly over the next 10-15 years driven by energy policy, entirely independently of AI.

If you are in a high-exposure role - clerical, administrative, data entry, back-office processing - and you are considering what to do about it: trades retraining is a structurally sound option that deserves serious consideration. It is not a consolation prize. It is a move from a role scoring 8.5/10 on AI exposure and facing 1-3 year disruption timelines (in high-velocity countries) to a role scoring 2.0-2.5/10 on AI exposure and in structural shortage with rising wages. The governments in the UK, Germany, and the US all fund trades retraining programmes specifically because the shortage is large enough that public investment in supply is justified.

The structural argument is this: in a labour market where AI is reliably compressing wages and reducing headcount in high-exposure professional and clerical roles, the trades represent a category of work where the opposite dynamics apply. Shortage is increasing wages. AI is not a threat. Demand is structurally rising. Those three things being true simultaneously, across 33 countries, is a signal worth taking seriously.