Software testers, writers and data-entry workers are among those seeing the biggest impact from AI, while construction and other hands-on occupations remain far less exposed, according to new analysis.
Computer and mathematical occupations currently show the greatest combined impact from AI automation and augmentation, followed by office and administrative support and sales roles, according to the American Staffing Association’s (ASA) AI Impact Index. At the other end of the scale, the index shows that construction and extraction, farming, fishing and forestry, and production occupations are seeing the least impact so far.
The ASA index combines AI usage data from Anthropic with occupational task data from the US Department of Labor’s O*NET database to assess where artificial intelligence is already changing work. Rather than predicting which jobs will disappear, it looks at which individual tasks within occupations are currently being automated or augmented by AI. The findings reveal a clear divide between digitally based, information-heavy work and many occupations that depend more heavily on physical skills and human interaction.
ROLES MOST AFFECTED BY AI
At individual occupation level, the ASA index identifies the three roles showing the greatest combined impact as:
- software quality assurance analysts and testers;
- writers and authors; and
- data-entry keyers.
By contrast, the index identifies 183 occupations where no AI automation or augmentation was detected at this stage, including 43 within construction and extraction. That does not mean those jobs will remain untouched by technology indefinitely. The index provides a snapshot of where AI is having an observable impact now, based on available Anthropic and O*NET data, rather than a forecast of future job losses.
WHITE-COLLAR WORKERS FEELING THE MOST IMPACT
The ASA findings suggest jobs built around digital information, writing, analysis and administrative processing are currently more exposed to AI than many occupations requiring physical presence, dexterity or hands-on skills. That helps explain why computer and mathematical occupations, office support and sales rank highly in the index, while construction and farming remain much less affected.
The findings also provide an interesting backdrop to growing concern among younger workers about traditional white-collar career paths. Eight in 10 Gen Z workers recently said they fear AI will replace traditional entry-level jobs, while almost half were either learning a skilled trade or seriously considering one. The same research found 82% believed skilled trades offer better long-term job security than many office-based careers.
The ASA findings do not prove that construction or other skilled trades are immune from future automation. But they do suggest that many hands-on occupations currently remain substantially less exposed to AI than information-heavy roles.
AI IS CHANGING TASKS BEFORE JOBS DISAPPEAR
A key feature of the ASA analysis is its focus on tasks rather than simply job titles. Jobs are made up of many different activities. Some may be automated almost entirely. Others may be carried out more quickly with AI. And many still depend heavily on human judgement, accountability, relationships or physical capability.
ASA measures the extent to which tasks are being automated or augmented and weights them according to their importance within each occupation. That means an occupation can register significant AI impact without the job itself disappearing. For employers, the more useful question may therefore be less: “Will AI replace this job?” and more: “Which parts of this job are changing – and what should people be doing instead?”
“In a sea of AI predictions, the ASA Impact Index allows hiring professionals a timely forecast of changes in labour demand based on what AI is taking off the plate of human workers,” said Noah Yosif, Chief Economist at the American Staffing Association. “Staffing and recruiting professionals can use the ASA Impact Index to anticipate client needs and position themselves as true workforce solutions partners.”
The ASA analysis suggests employers may need to look more closely at what is changing inside jobs, rather than relying on broad labels such as “AI-proof” or “at risk”.
PRODUCTIVITY GAINS DON’T NECESSARILY MEAN LESS WORK
The ASA findings also fit with wider evidence showing that AI is changing the way employees work even where job titles and headcounts remain unchanged. Seven in 10 employees already say AI makes them feel more productive, yet workloads are not falling.
AI can reduce time spent drafting, summarising, analysing or processing information, but it can also create new work around checking outputs, correcting errors, supplying context and making decisions. The result may therefore be job transformation rather than immediate job disappearance. For employers, that raises another question: when AI saves time, is that capacity being used to improve work, or simply to increase workload?
COULD AUTOMATION REMOVE THE TASKS PEOPLE LEARN FROM?
There is another potential consequence for employers. Some of the routine tasks AI can perform most easily are the same activities through which less experienced workers traditionally develop their skills.
Junior employees often build expertise by carrying out relatively routine work, making mistakes, receiving feedback and gradually taking on more complex responsibilities. If AI increasingly performs that work, organisations may need to deliberately create alternative ways for people to build judgement and experience. Six in 10 workers already worry AI is weakening their skills, while wider research suggests critical thinking and problem framing are becoming increasingly valuable as AI takes on more routine cognitive work.
The concern is therefore not only whether AI removes junior jobs. It is whether organisations automate the very experiences through which tomorrow’s experts and leaders learn. Businesses could successfully automate today’s repetitive work while unintentionally weakening their future talent pipeline.
WHY AUTOMATION DOESN’T NECESSARILY MEAN REPLACEMENT
The ASA index measures both automation and augmentation. The distinction matters. Automation means AI carries out an activity previously performed by a person. Augmentation means AI helps somebody complete that task.
A writer, for example, may use AI to summarise material or produce an initial draft while remaining responsible for accuracy, judgement, tone and final output. A software tester may use AI to generate test cases or identify potential faults while humans continue making higher-level decisions about quality and risk. An administrative employee may automate document preparation but spend more time handling complex requests, relationships or coordination.
So AI affecting a task does not automatically mean the person performing the job becomes unnecessary.
HUMAN JUDGEMENT COULD BECOME MORE VALUABLE
As AI takes on more routine work, some human capabilities may become more important rather than less. Critical thinking, judgement, communication, creativity, accountability and relationship-building become increasingly valuable when employees are expected to oversee AI rather than simply carry out the underlying work themselves.
But human oversight only works if people retain the ability and confidence to challenge what technology produces. Four in 10 US workers using AI recently admitted accepting an answer they suspected was wrong, while only 35% said they always independently verified AI outputs. The future of work therefore cannot simply be about teaching employees how to use AI. It also means ensuring people retain the expertise needed to recognise when AI is wrong.
WHAT EMPLOYERS SHOULD DO
The ASA findings suggest broad predictions about whether an entire occupation is “safe” or “at risk” may be less useful than examining how individual tasks are changing. Employers should:
- Break jobs down into tasks rather than treating entire roles as AI-exposed.
- Identify which activities can be automated, augmented or should remain human-led.
- Protect work requiring judgement, relationships, accountability, creativity or ethical decision-making.
- Invest in reskilling before displacement occurs.
- Monitor whether employees are losing important capabilities because AI increasingly performs tasks for them.
- Redesign jobs around higher-value work rather than simply increasing workloads when AI saves time.
- Review entry-level positions carefully so automation does not remove the experiences through which people learn.
- Give employees opportunities to practise decision-making, problem-solving and independent thinking.
- Involve HR and workforce leaders early when deciding which tasks should move to AI.
- Measure whether automation is genuinely improving jobs and productivity rather than simply increasing output expectations.
Strong AI governance also appears to have commercial benefits. Organisations with trustworthy AI practices are 15 times more likely to report strong returns from AI, reinforcing the case for accountability, data quality, explainability and human oversight alongside adoption.
WHAT WORKERS CAN DO
For workers, the findings provide a more useful way to think about AI risk than simply asking whether technology can “take my job”. Instead, employees can ask:
- Which of my regular tasks could AI already perform?
- Which activities require contextual knowledge or judgement?
- Where do clients, customers or colleagues value human interaction?
- Which parts of my role involve responsibility that cannot easily be delegated to technology?
- What adjacent skills would make me more valuable if routine work disappeared?
- How can I use AI to improve my work rather than compete directly with it?
Becoming skilled at both using AI and supervising it may ultimately prove more valuable than trying to avoid the technology altogether. That applies even to many hands-on occupations. The recent Gen Z skilled-trades research found AI is already being used by tradespeople for activities such as marketing, customer communications and research, even where core physical work remains substantially less exposed.
HOW AI WILL IMPACT JOBS
Much of the debate around workplace AI has focused on one question: Which jobs will disappear? The ASA AI Impact Index suggests that may be too simplistic. Jobs are collections of tasks. AI may automate some, augment others and leave many largely untouched. As that combination changes, occupations themselves can change significantly without disappearing altogether.
The bigger workforce challenge may therefore be less about predicting precisely which jobs survive and more about ensuring employers understand which work should move to machines, which should remain human and how people continue developing the skills needed for what comes next.
For workers, the same principle applies. The safest career may not necessarily be one completely untouched by AI. It may be one where people know how to use the technology, while continuing to build the judgement, expertise, creativity and relationships that remain much harder to automate.
Click here for more information about the ASA Index.
RELATED ARTICLES
Six in 10 employees worry AI is weakening their skills, while 43% say they get the blame when AI goes wrong, new global research reveals.
The majority of young Americans believe AI will eliminate many traditional entry-level jobs, so 92% have changed their career strategy and almost half are seriously considering a skilled trade, new research reveals.
Seven in 10 employees say AI is making them more productive, yet workloads remain virtually unchanged, reveals new research.
More than four in 10 US workers surveyed who use artificial intelligence (AI) at work have knowingly accepted an AI-generated answer they suspected was wrong because it was faster or sounded convincing, according to new research.
Organisations with strong AI governance, data quality, explainability and accountability are 15 times more likely to report strong returns from artificial intelligence investments, according to new global research.
