Geoff Crittenden, CEO of Weld Australia, says that welding remains a future-proof career despite AI disruption.
If you believed every headline over the past year, you would think half the workforce is about to disappear.
Lawyers, coders, analysts, marketers, administrators. Every week there is a new prediction about how artificial intelligence will wipe out white-collar work as we know it. Some of that concern is real. Some of it is overblown. But one thing is increasingly clear: the jobs most exposed to AI are not necessarily the ones many people assumed would be first in the firing line.
In fact, some of the jobs with the strongest future are the ones that require human judgement, physical skill, adaptability, problem-solving and the ability to work in messy, unpredictable real-world environments. In other words: the trades.
We have been here before. The first great digital disruption came with the arrival of desktop computers and PCs, and it changed the workforce almost overnight. Typing pools vanished, many administrative roles were absorbed into software, and work that once required entire back-office teams was redistributed or eliminated. That was the first wave. AI is the second. But while technology has steadily replaced routine clerical work, the demand for welders and tradespeople has moved in the opposite direction. It has grown, and it continues to grow. That is because welding, fabrication and the skilled trades sit in the physical economy, where judgement, dexterity, adaptation and hands-on problem-solving still matter.
Look at the Australian data. PwC’s 2025 AI Jobs Barometer found a strong negative relationship between AI exposure and growth in jobs in Australia. The jobs least exposed to AI recorded average growth of 43 per cent to 44 per cent, while the jobs most exposed grew by just 10 per cent to 14 per cent. Importantly, PwC’s Australian analysis places building and related trades workers among the less AI-exposed occupations. In other words, the closer a job is to screens, forms, routine analysis and repeatable information work, the more exposed it appears to be. The closer it is to physical work, variable environments and hands-on judgement, the more resilient it looks.
And this is no longer just theoretical. In March 2026, Atlassian announced 1,600 job cuts globally, including about 480 roles in Australia, with the company saying the savings would be used to fund further investment in AI and enterprise sales. That is one of the clearest local examples yet that AI disruption is hitting digital and knowledge-based work first. It is not welders being laid off because a robot can suddenly crawl inside a ship hull and perform coded welding in a live industrial environment. It is office-based, screen-based and software-driven roles that are being restructured first.
That should make all of us stop and think.
For years, the assumption was that trades would be the jobs disrupted by automation while office-based knowledge work would remain secure. AI has turned that logic upside down. It is proving good at generating text, summarising information, writing code, creating presentations and even producing legal arguments.
In Australia and overseas, we are already seeing what happens when people trust those tools too much. Earlier this year, India’s Supreme Court warned that reliance on AI-generated fake judgments would amount to misconduct after a lower court relied on non-existent case law.
That is a pretty good reminder that AI can sound confident without being correct. Now compare that to welding.
A welder has to interpret drawings, understand fit-up, work with different materials, manage distortion, respond to changing conditions, assess a joint in context and physically execute high-quality work in environments that can be cramped, dirty, elevated, hazardous or highly variable. Anyone who thinks a chatbot is about to replace that has probably never been inside the hull of a ship, underneath a power station, or on a live construction or shutdown site.
Yes, robotics and automation already play a role in welding. They should. They improve productivity, consistency and safety in the right applications. But that is not the same as replacing welders. It is about changing the mix of work. Automation is effective in controlled, repeatable manufacturing environments. It is less effective when the work is complex, variable, high-consequence or physically constrained. Building a humanoid robot is one thing. Getting it to perform coded welding inside a confined industrial environment, with real-time judgement and accountability, is another thing entirely.
Welding is a great career
That is why I believe welding has always been a great career, and will remain one well into the future.
The labour market backs that up. Jobs and Skills Australia said technicians and trades workers are projected to grow by 195,800 over the decade to 2035. At the same time, the Occupation Shortage List shows that nearly half of trade roles remain in shortage, even though overall labour market pressures have eased.
Welders are part of that picture. Jobs and Skills Australia’s latest occupation profile shows there are around 19,900 Welders (First Class) employed in Australia, working primarily in manufacturing and construction. The same profile shows the occupation remains overwhelmingly male, with women making up just 1 per cent of the workforce, which tells us there is still huge room to grow the pipeline if we are serious about the future of the trade.
This is where the conversation about AI and trades needs to mature.
The real story is not “AI versus welders”. The real story is that AI will change how welders work, not whether welders are needed. It will help with documentation, quality systems, planning, knowledge retrieval, training support, simulation, defect analysis and workflow optimisation. It will make good welders more productive and good businesses more efficient. But it will not eliminate the need for skilled people who can build, repair, inspect and maintain physical assets.
And that matters for Australia.
The National AI Centre’s 2025 ecosystem report shows AI-related jobs and skills are in demand across the economy, and that Australia’s AI ecosystem includes both organisations developing AI products and businesses actively adopting AI in operations. We should absolutely embrace that. But embracing AI does not mean pretending every occupation will be transformed in the same way.
It also means recognising where real strategic vulnerabilities sit.
Australia still needs people who can build ships, fabricate pressure equipment, erect steel, weld transmission infrastructure, maintain processing plants, repair defence assets and deliver the energy transition. None of those things happen without trades. In fact, as more of the economy becomes digitised, the jobs that remain grounded in physical reality become more valuable, not less.
A career in trades is a smart long-term choice
That is why we should be saying this much more clearly to young people, parents, schools and governments: a trade is not a fallback option in the age of AI. It is one of the smartest long-term career choices available. If you want a career that is portable, practical, in demand, tied to real economic value and far less vulnerable to generative AI disruption than many office jobs, welding is a very good place to start.
But we cannot just say that. We have to back it.
That means revolutionising how we train the next generation of tradespeople. It means investing in school pathways, apprenticeships and TAFE. It means using better technology (like simulators and digital learning tools) to attract and train the next generation. It means lifting the status of trades as professions that require skill, precision, discipline and judgement. And it means making sure industry is part of the conversation about what future work looks like, rather than leaving that debate to people who have never set foot in a workshop.
A training model built around decades-old TAFE structures and three- to four-year apprenticeships is no longer agile enough for the workforce pressures we face today. Industry does not need young people spending years sweeping floors, grinding metal and making coffee before they are trusted with meaningful work.
Industry needs smarter, faster, more flexible pathways that bring apprentices onto the tools earlier and train them to internationally recognised standards linked to real welding procedures – not vague or outdated notions of units of competency. That is better for employers, but it is also better for apprentices. It gives apprentices clearer goals, faster progression, stronger qualifications and a real sense of value earlier in their careers. If we are serious about solving skills shortages, then we need training systems that get people job-ready, qualified and productive far sooner, because industry needs them now.
Because while the newspapers are full of stories about AI replacing workers, the bigger risk for Australia is something else entirely: that we fail to train enough tradespeople to build the future.
AI will write more documents. It will generate more reports. It will answer more emails.
But AI will not weld a pressure vessel to code, repair a fatigued structure in the field, or work out how to complete a critical fabrication job when the fit-up is imperfect and the clock is ticking.
A skilled welder can.
.jpg)
