Create a free account, or log in

Employment Hero debuts SmartMatch amid concerns over AI and recruitment bias

Employment Hero’s new AI tool, SmartMatch, aims to revolutionise SME recruitment but raises concerns about potential bias and fairness.
Tegan Jones
Tegan Jones
ben thompson employment hero smartmatch AI bias budget
Employment Hero CEO and co-founder Ben Thompson. Image: Supplied

Employment Hero has launched SmartMatch, an AI tool the company says is designed to revolutionise how SMEs connect with potential employees. However, this advancement comes amid growing concerns about AIโ€™s role in the job market, especially regarding bias and access to opportunities.

According to Employment Hero CEO and co-founder Ben Thompson, SmartMatch taps into the vast but often overlooked “hidden job market” by leveraging artificial intelligence to match job seekers with suitable roles, particularly in the small to medium business spaces. This is based on their skills and career aspirations, without relying on traditional job advertisements.

โ€œWith SmartMatch, job seekers are able to access a greater pool of jobs and even apply for opportunities that may not be listed elsewhere,โ€ Thompson said to SmartCompany.

SmartMatch analyses real-time data from both job seekers and employers, facilitating instant connections and aiming to streamline the hiring process.

Employment Hero says this approach not only cuts down on recruitment time and costs but also aims to enhance accessibility and equity in the job market.

โ€œEmployment Heroโ€™s own research has shown that 75% of SMEs struggle to compete for talent against big businesses,โ€ Thompson said.

โ€œIn addition, traditional recruitment methods are costly and slow — the average cost to hire someone regularly exceeds $5,000.โ€

Addressing bias concerns in AI recruitment

Despite these intentions, scepticism remains regarding AI in recruitment, particularly concerning fairness and transparency.

A 2023 report by Roy Morgan highlighted that a significant portion of Australians, especially women and rural residents, are wary of AI, fearing it might exacerbate existing problems rather than solve them. This concern is not unfounded, as AI systems, including those in recruitment, have historically shown tendencies to perpetuate existing biases unless explicitly corrected.

This is largely due to them being informed by inherent bias, as well as trained on global data sets that lack diversity.

One of the most contentious points surrounding SmartMatch, and similar AI-driven tools, is the assertion that they can help employers find candidates who are “identical” or highly similar to their current successful employees.

While this may sound beneficial for maintaining a certain standard within the workforce, one could argue that this could inadvertently reinforce existing workplace homogeneity, effectively sidelining diversity.

This propensity for bias stems from the risk that AI systems when trained on historical employment data, might perpetuate the preferences and prejudices implicitly contained within that data.

For instance, if a company’s historical hiring patterns show a tendency to select candidates from a specific demographic or educational background, the AI might infer these traits as preferred, potentially disadvantaging equally capable candidates who differ from that.

In response to these concerns, Employment Hero says it has taken proactive steps to mitigate the risk of bias.

“The AI is trained to analyse the skills and experience of candidates in the system in order to present employers with staff that best fit their hiring needs,โ€ Thompson said.

โ€œNo personal or sensitive information is included as inputs into the recommendation engine to minimise the risk of introduced bias.

“The team uses several techniques to mitigate bias including, controlling data sets, having transparency on data factors used, and in some cases positioning AI as assistive for human made decisions.”

Further addressing the challenges of AI fairness, Thompson noted that more work is being done, saying that SmartMatchโ€™s algorithms are constantly being refined and adjusted.

โ€œSmartMatch is still in its infancy and we are working incredibly close with our users, both on the employer and candidate side to ensure the best quality matches,โ€ Thompson said.

โ€œAs part of this and as we continue to build out SmartMatch, we intend to build out bias measurement tools to ensure that the system doesn’t skew towards any particular demographic.โ€

Thompson also revealed Employment Hero has built an evaluation framework to measure the performance and accuracy of SmartMatchโ€™s recommendations.

โ€œThis is something we intend to invest heavily in, in order to ensure the recommendations provided by SmartMatch are as relevant as possible,โ€ Thompson said.

Interestingly, AI might also offer a solution to gender disparities in recruitment. According to a white paper on AI and gender diversity by Sapia.ai in 2023, women are 30% more likely to complete a job application if the process involves AI, possibly due to perceived lower bias compared to human evaluators.

This suggests that when properly managed, AI could indeed play a crucial role in creating a more equitable job market.

However, the technology’s potential to disrupt traditional job platforms poses another layer of complexity. With SmartMatch, Employment Hero is not just challenging the status quo of job advertising but also the economic model that underpins much of the recruitment industry.

By bypassing job boards and directly connecting employers with potential employees, SmartMatch could significantly affect job market dynamics, potentially sidelining established players like Seek.

In his vision for SmartMatch, Thompson envisions a streamlined job market where “talent flows like water through a pipe,” removing what he describes as a “handbrake on the global economy” due to inefficient hiring practices.

Yet, as with all new technologies, the broader implications of such disruption will need careful monitoring to ensure that the benefits are evenly distributed across all sectors of the job market.