Right now, there is a quiet war playing out inside your hiring pipeline. Candidates are using AI to mass-apply for jobs. Employers are using AI to mass-screen those applications back out, just as fast. Two automated systems negotiating with each other – while the human on either side gets less of a say by the month.
Every enterprise HR leader and government procurement manager is caught in the middle of it, under the same pressure: process more applications in less time, while still hitting ambitious diversity, equity and Reconciliation Action Plan (RAP) targets.
On paper, automated resume screening looks like the answer to both problems at once. In practice, the data says otherwise – it’s creating a silent, costly failure inside the exact pipeline it was built to fix.
The Myth of the Neutral Algorithm
Resume-screening algorithms are built to spot patterns in past hiring data. If your historical workforce lacks diversity, the software quietly learns to reward candidates who resemble the people you’ve already hired, and to screen out anyone who doesn’t fit that template.
In the most detailed study of its kind, a joint Harvard Business School and Accenture survey of 2,275 executives and 8,720 workers across the US, UK and Germany – 88% of employers admitted their own automated systems filter out qualified, high-skilled candidates simply because a resume doesn’t match the exact wording of the job description. For middle-skilled roles, that figure rises to 94%. Employers in the same study estimated that filtering alone eliminates more than half of all candidates for a role before a human ever sees the shortlist.
In an Australian context, with RAP commitments on the line, the same mechanism has a sharper edge. A candidate with rich community leadership experience, a non-linear career path, or a resume that doesn’t use the same job titles as your last ten hires isn’t read by the algorithm as differently qualified – they’re read as a data mismatch, and dropped in seconds.
Three Hidden Risks of Bot-Led Hiring
- Breached RAP commitments. Almost half of employers (48%) configure their systems to automatically screen out any candidate with an employment gap of six months or more – a filter with nothing to do with capability, and everything to do with penalising exactly the kind of career break common among carers, community leaders, and First Nations candidates balancing cultural responsibilities. A RAP built on genuine intent can be quietly undone at the ATS configuration screen, with no one in the business aware it’s happening.
- Talent monocultures. A system built to reward exact keyword and title matches doesn’t reward adaptability, problem-solving, or lived experience — it rewards whoever is best at describing themselves in the software’s preferred language. That’s a different skill to the one you’re actually hiring for, and it’s how organisations end up with a workforce that looks, on paper, more alike than they intended.
- Brand and candidate erosion. Recent research suggests around two in three job seekers now assume they’ve been rejected by a machine rather than read by a person. Every strong candidate who disappears into that black hole without a response takes that perception with them — whether or not your organisation is actually doing anything wrong.
Where Automation Actually Belongs
To be clear, this isn’t an argument against AI in hiring. Used well, automation is exactly what lets a lean HR or People & Culture team handle a volume of applications it could never manage by hand. And it’s worth noting that even recruiters themselves don’t fully trust these tools blindly: most treat an AI match score as guidance to be verified, not a definitive filter. If the people running the system don’t trust it to make the call alone, it’s a fair question why the process still lets it make the first cut alone.
Automation belongs in backend operational workflows – scheduling interviews, managing data flow, syncing candidate records between platforms. It does not belong in the room evaluating human potential, cultural fit, or the kind of complementary team skills no keyword field will ever capture. That evaluation is, and should stay, a human judgement call.
| Approach | What it optimises for | What it actually delivers |
|---|---|---|
| Traditional hiring | Manual review at every stage | Thorough, but slow – heavy admin overhead and long time-to-hire |
| AI-only screening | Speed and volume | Fast, but blind – silent diversity bias and a shrinking, monocultural shortlist |
| Goanna’s human-first model | Automation for admin, humans for judgement | Fast where it can be, careful where it matters – operational efficiency and authentic merit |
The Bottom Line
True workforce alignment – the kind that actually moves a RAP scorecard, not just a hiring KPI – still requires a human being to make the final call on human potential. The question isn’t whether automation belongs in your hiring process. It’s whether you actually know where the line between automation and evaluation currently sits in yours.
Sources: Harvard Business School & Accenture, “Hidden Workers: Untapped Talent” (2021), surveying 2,275 executives and 8,720 workers across the US, UK and Germany; Reconciliation Australia, 2025 RAP Impact Report; Enhancv, 2025 job-seeker perception survey.