Frontline workers use AI, but employers lag on guidance

Humanforce research found that 65% of Australian frontline workers have used AI or an AI-enabled tool at work in the past year, while 46% had received no employer training, briefing or written rules on AI.

The findings are based on a survey of 501 workers in healthcare, hospitality and tourism, and retail and wholesale trade. Among those who had used tools such as ChatGPT, Gemini, Claude or Copilot for work, 24% said their employer did not allow it or they did not know whether it was allowed.

That figure rose to 36% in healthcare. Across the survey, only 13% of workers said they had received comprehensive training.

The results point to a gap between the spread of AI in frontline jobs and the guidance employers provide. Workers were most interested in using technology to reduce routine administrative work.

Paperwork and administration ranked first in every sector as the area workers most wanted technology or AI to make easier, cited by 43% overall. Rostering and shift management followed at 31%.

Nearly half of respondents, 49%, said they spend at least two hours a week on tasks such as forms, timesheets, shift changes, finding information, and resolving pay or leave issues. More than one in four, 28%, said they spend four hours or more on those tasks.

"Australia's AI debate still revolves around people working at desks. Yet frontline workers are already using AI on the job, and guidance hasn't kept pace," said Clayton Pyne, Chief Executive Officer of Humanforce.

"Someone caring for patients, serving customers or moving between shifts should know what they can share with general-purpose AI tools, what they can rely on, and when an answer needs checking. Organisations that provide this clarity early will reduce risk and get far more from AI, including less paperwork, easier shift management and more time with customers and patients," Pyne said.

Pay concerns

The survey also highlighted persistent payroll problems in shift-based work. Two in five workers, 41%, said their pay had been wrong at least once in the past 12 months, while 27% said it had happened two or more times.

Humanforce linked this to the complexity of awards, penalty rates and roster changes in frontline industries, which can make errors harder to detect and correct.

Hospitality and tourism workers reported the highest rate of pay errors, with 48% saying they had been paid incorrectly in the past year, compared with 33% in retail and wholesale.

"Frontline workers feel that admin burden most directly, but for organisations it is also a compliance, cost and productivity issue, which makes the back office a place where AI can provide significant support," Pyne said.

Pyne also pointed to the need for safeguards around the use of AI in payroll and compliance processes, saying the technology should operate within clear permissions and with traceable, explainable actions.

"Getting pay right is one of the most basic commitments an employer makes, and one of the hardest to keep in shift-based work, where errors are easy to make and hard to spot," he said.

"AI will do far more to help, but only when it is built on the rules, data and approvals that already make pay compliant. Nobody should have to take AI's word for what they're owed," Pyne said.

Human contact

Workers also identified training and support, workload and wellbeing, and understanding pay and entitlements as areas where they wanted more help from technology. Training and support was cited by 27%, managing workload, fatigue or wellbeing by 23%, and understanding pay, entitlements or awards by 19%.

At the same time, respondents expressed concern about AI's effect on the human aspects of frontline work. Losing human contact was the top concern overall, selected by 47% of workers.

Healthcare workers reported the heaviest administrative load, with 58% saying they spend two hours or more each week on admin. Half of healthcare respondents said they had received no guidance from their employer on AI, and misuse of personal information was their top concern at 40%.

Retail and wholesale workers were the most likely to have received some form of AI guidance, although 37% still said they had received none. Three in 10, 30%, said they use AI most days, while 53% identified losing human contact as their main concern.

Support for AI use appeared to depend on retaining human oversight. Almost half of respondents, 49%, said the most important safeguard was having a person review significant AI decisions, while 39% wanted the ability to question or appeal an AI-assisted decision. Only 8% said no additional safeguards were needed.

"Much of frontline work is human by nature: empathy, in-person care and service, from a carer helping a resident start their day to a host welcoming a guest. That helps explain why paperwork topped their list," Pyne said.

"Workers want technology to take on the admin so they have more time for the people in front of them," he said.

"The frontline organisations that get the most from AI will use it to give their people more time for the human side of the job," Pyne said.

"Workers should never be left arguing with an AI system about a decision that affects their hours, pay or working life. Organisations need to be clear about where AI is used, keep people involved in important decisions and give workers a way to speak up when something looks wrong," he said.