The rights of employees must be considered in the rollout of the technology beyond job security and training, workplace standards expert Penny Williams told a parliamentary inquiry into AI.
"Various forms of AI have been introduced to the benefits of workers, but it can also be used punitively," the Queensland University of Technology professor said.
"It isn't a straight line to say it's the developer's responsibility, or the technology itself and the algorithm itself that is the issue."
Representatives from multinational Amazon faced questioning from a Labor senator over documents showing the extent of surveillance in its workplace.
Internal documents leaked from the Canadian branch of Amazon revealed instances of employee dissatisfaction were being tracked and measured across sorting facilities, including one in Melbourne.
The so-called Atlas Model operated by Amazon tracks employee relations events internally as well as external comments on websites such as Reddit to "evaluate emerging workforce issues" according to the training document.
It includes a breakdown chart analysing global warehouse trends, noting 19 external comments were recorded in relation to the Melbourne fulfilment centre with 80 per cent classed as negative.
Constant surveillance could be dangerous in the workplace, occupational health and safety expert Michael Quinlan said.
"If you're tracking people and they're not aware of it, or even if they are aware of it ... this is going to affect their behaviour," Professor Quinlan from the University of NSW told the inquiry in Canberra.
"For instance, some of the surveillance equipment on trucks on mining sites actually gets drivers to look in a particular direction all the time and they're arguing that is actually distracting because they know they're being recorded."
Labor Senator Tony Sheldon pushed Amazon Web Services representatives on the use of AI-powered tracking tools.
Amazon Web Services is the retail and tech giant's dog in the AI fight, which essentially functions as a utility company - like electricity and water - but for computing power.
Representatives from the services team said they were unable to comment, having only been made aware of the report moments before fronting the inquiry.
Amazon executive Nicole Foster instead spoke to the standards applied to responsible use of their services more broadly.
"We use responsible AI dimensions that are more around technical language ... safety, veracity, controllability, fairness, explainability, robustness, privacy and security," she told the inquiry.
But Senator Sheldon suggested the company was failing to apply the same standards to its own use of AI in the workplace.
"There is not privacy," Senator Sheldon told the inquiry.
"Workers have not been told that they are being surveilled.
"They've been treated like a digital prison."
The NSW senator raised particular concern with cases in which people have been terminated from their jobs for raising concerns or joining a union.
"It's a bit like they're using AI to hunt down the workforce that are pursuing their human rights," Senator Sheldon said.
"It's a bit like the workforce being hunted down by the Terminator."
Amazon Australia has been contacted for comment.
Representatives before the inquiry welcomed calls for mandatory safety standards and working toward supporting renewable energy projects to meet data centre energy needs.
However, Amazon dismissed calls to increase the resources allocated to the Australian AI Safety Institute, claiming was a matter of "quality over quantity" at such an early stage of the body's existence.
The tech giant has labelled the artificial intelligence boom the first real opportunity to boost global productivity growth in 20 years, echoing analysis from Australia's Productivity Commission.
20°C