A report released by The University of Technology Sydney on October 20 calls for stronger protection against technology-enabled workplace surveillance.

Subscribe now for unlimited access.

Login or signup to continue reading

Canberra should take the warning seriously.

Its authors propose a fair and reasonable test for monitoring and a dedicated complaints pathway through the Fair Work Commission. When an automated system flags a public servant for poor performance, who checks whether the allegation is fair before it damages a career?

The Australian Public Service has a starting point.

The APS Commission's September consultation guidance requires agencies to consider consultation with employees and unions when proposed changes materially affect their work. It distinguishes incidental AI features from major changes to roles and practices. Automated performance monitoring deserves scrutiny before an agency purchases software or starts collecting data, but advance consultation cannot resolve every mistake made after deployment.

Imagine a case officer who spends additional time checking contradictory evidence before deciding an application. An AI-enabled dashboard compares that employee with colleagues processing straightforward files and labels the officer inefficient. The supervisor sees a red performance indicator, without knowing that the slower employee may be preventing costly errors. Faster decisions become a target even when careful decisions better serve the public.

That hypothetical is especially relevant to a public service whose work ranges from routine processing to sensitive assessments. A numerical measure can flatten differences in case complexity, collaboration and professional judgement. Once the measure feeds performance discussions, employees have an incentive to satisfy the dashboard.

The agency may discover too late that it has rewarded shortcuts and discouraged staff from raising difficult questions.

Every Commonwealth agency using algorithmic monitoring for employment decisions should adopt a clear review rule. An automated score or alert should never, by itself, trigger a formal warning, adverse........

© Canberra Times