Employers in Australia are increasingly employing artificial intelligence and monitoring tools to monitor their employees, often without the knowledge of the employees. While a number of existing protections exist for the workplace, the report from the Human Technology Institute at the University of Technology Sydney (UTS) suggests they have not kept up with these technologies.
The issue of security cameras is not the only concern. AI systems can track employee activity, alert them to errors, and provide insights into computer usage, aiding employers in performance evaluation. As the use of these tools increases, it poses tough questions of workers’ privacy, dignity and decisions based on automated monitoring that researchers say is not fair, the researchers said.
From CCTV Cameras to AI That Flags Tiny Mistakes
Drawing the Line is published by the University of Technology Sydney’s Human Technology Institute. It explains the use of monitoring in retail stores, factories, delivery work to offices, and the expansion of monitoring.
There are multiple methods that an employer can use to monitor their employees with the use of technology, such as:
- Use of CCTV cameras to monitor workplaces.
- The monitoring of keystrokes that captures the keystrokes made by employees while working on the computer.
- Systems for monitoring messages and computer use at the job.
- Records that can be used to monitor employee movements via the use of cards.
- Patterns and/or indicators of potential performance issues from AI systems.
There are some of these tools have a valid use, like in workplace safety and security. Concerns include the possibility that employers gather too much information, don’t inform employees of what they are monitoring, or overuse automated measurements.
One case in particular was reported by The Guardian: a supermarket cashier was warned in writing after committing too many mistakes, which were picked up by a computer system overseeing their work. The errors reported were 0.18%, which translates to about 1 error per 533 items scanned. It was referred to a Parliamentary inquiry on AI in a submission by Australia’s Shop, Distributive and Allied Employees Association.
That case serves as a reminder of a possible issue when trying to automate performance tracking. A system may detect an error, but this does not necessarily prescribe the cause of an error or whether or not a disciplinary action is fair.
Researchers Want Stronger Rules for AI Workplace Monitoring
The Human Technology Institute is seeking a change to the Fair Work Act to clearly protect against detrimental workplace surveillance.
The main principle is that employee monitoring should be “fair and reasonable,” affording employers the flexibility to monitor employees if there is a legitimate business need, but not too much or too little monitoring.
The report also makes a number of recommendations:
- Informing employers of the monitoring systems they plan to implement, and involving workers in the process.
- Providing a special channel for staff to submit complaints regarding matters of privacy to the Fair Work Commission.
- Offering regulators a clearer and more concise idea of acceptable workplace monitoring.
- Permitting ministers to block surveillance products which can cause unacceptable harm to workers’ rights.
These are recommendations, rather than a description of protections which are already in full effect. The institute says that more clear-cut regulations might help employees to better understand their rights and provide greater clarity for employers.
The Next Challenge Is Finding the Line Between Oversight and Privacy
There are good reasons why employers should monitor some work-related activities. A delivery company might need to monitor vehicles for safety reasons, or a business could monitor computer activity to follow up on an incident of security.
Monitoring becomes more problematic if employees aren’t sure what is being monitored; if private activity is recorded where it doesn’t need to be recorded; or if a disciplinary measure is taken for private activity that had been automatically monitored but not adequately reviewed by humans.
The challenge is to get AI to have a positive impact on the workplace without making everything an evaluation score.
It may be interesting to see how the Australian government reacts to the institute’s recommendations. The report comes at a time when Parliament is considering AI in the workplace in Australia.
The question for workers is simple: They shouldn’t give up any right to privacy while they’re at work. As the monitoring of employees becomes more effortless and detailed with the aid of AI, transparency, reasonable limitations, and human oversight will be more essential.
