Merit protection commissioner releases new guidance on AI recruitment

article-body

New guidance has been released for APS agencies by the merit protection commissioner on the use of AI-assisted recruitment processes.

The guidelines cautioned agencies on the use of AI in recruitment, reminding the APS the merit principle is still upheld when using AI-assisted tools in recruitment.

“There should be a clear demonstrated connection between the candidate’s qualities being assessed and the qualities required to perform the duties of the job,” the guidance stated.

The guidance has been published following the news in the MPC’s annual report for 2021-22 found a number of recruitment decisions had to be overturned.

Then commissioner Linda Waugh stressed the importance of the merit principle, as previously reported by The Mandarin.

“Merit should not only be viewed as applying to recruitment. It should be the underpinning basis of all government decision-making, including decisions relating to our most valuable asset: our APS employees,” Waugh said at the time.

Three myths were debunked as part of the new guidance.

The first was that all AI-assisted tools were thoroughly tested, with the guidance pointing out that there was limited guidance on these tools and the quality of AI-assisted tools varies greatly.

Another was the reproduction of bias within AI, with the guidance stating AI can have bias.

“For example, if an AI is based on a dataset of interviews where all successful candidates cough during the interview, the AI may determine that coughing makes a candidate suitable for the position,” the guidance read.

The third myth the guidance sought to debunk was that agencies were not responsible for decisions an AI makes — agencies “are accountable for ensuring their recruitment processes follow the merit principle”.

The guidance listed out some of the benefits of using AI recruitment tools, including a reduction in resourcing of recruitment, reduction in some bias compared to traditional recruiting, consistency, and improvement of candidate’s experience.

On the other side, the risks of using AI “incorrectly” included assessing candidates on something other than merit, ethical and legal concerns about transparency and data privacy, and biased results from either poor data or algorithmic bias.

---

:

Third-party AI throws spanner in works for fair APS recruitment process