The Best Candidate You Never Saw

A woman interviewed on the Today Show recently shared an experience that should make every employer stop and think. 

She was over 50, highly qualified and repeatedly applying for roles she was well equipped to do. Yet she could not get past the first stage. No interview. No conversation. Just rejection. 

Then she changed her resume. She removed the dates that allowed an employer—or its screening technology—to infer her age. The substance of her capability did not change. Her experience did not change. Her potential contribution did not change. But suddenly, she began securing interviews and ultimately landed a job. 

That experience does not, on its own, prove precisely how every application was assessed. But it exposes a risk organisations cannot afford to dismiss: the technology designed to help you identify talent may be screening out some of your strongest candidates before a human ever sees them. 

Efficiency is not the same as effectiveness 

AI can bring real value to recruitment. It can process large volumes of applications, identify relevant skills, streamline administration and reduce the time it takes to create a shortlist. 

But speed is only useful if the process is producing the right outcome. 

A recruitment system that rejects excellent candidates more efficiently is not a better system. It is simply a faster way to make a poor decision. 

The danger comes when employers assume that because a process is automated, it is objective. AI does not arrive free from bias. It learns from data, criteria and patterns shaped by people. If historical hiring decisions favoured particular ages, backgrounds, career paths or styles of communication, an AI system may learn to reproduce those preferences—and do so at a scale no individual recruiter ever could. 

What is your system really selecting for? 

Most organisations intend to screen for capability, potential and alignment with the role. But without careful oversight, automated tools may place weight on factors that are only proxies for those things. 

Dates of graduation may signal age. An extensive career history may be interpreted as overqualification. A period out of the workforce may count against someone who took time away to care for children, ageing parents or their own health. Language, accent, disability or a non-linear career path may influence how a candidate is assessed even when none of those factors affects their ability to perform the role. 

In each case, the system may appear to be assessing merit while quietly narrowing the field to people who resemble those hired before. 

That does not just create a fairness problem. It creates a talent problem. 

The commercial cost of invisible candidates 

The candidate screened out at the first stage may be the person with the judgement to prevent an expensive mistake, the resilience to lead through uncertainty, or the depth of experience to coach less experienced colleagues. 

They may also bring perspectives your existing workforce lacks. Yet because they never reach interview, leaders do not know what they have missed. There is no obvious vacancy in the shortlist labelled 'experienced candidate excluded by flawed assumptions'. The loss is invisible. 

This is why AI screening is not merely a technology or recruitment issue. It is a talent strategy issue. If your stated strategy is to build a diverse, capable and adaptable workforce, but your screening process rewards conventional career histories and filters out difference, the process is working against the strategy. 

Human oversight must mean more than clicking approve 

Keeping a person somewhere in the process is not enough. If that person only reviews the candidates the system has already selected, they cannot challenge the quality of those it rejected. 

Meaningful oversight requires employers to understand how the tool works, what information it uses, what it has been trained to value and whether particular groups are being disproportionately excluded. 

At a minimum, employers should: 

be clear about where AI or automation is used across the recruitment process; 

understand and challenge the criteria used to rank, progress or reject candidates; 

test outcomes for patterns of exclusion across age, gender, disability, cultural background and other protected attributes; 

regularly review a sample of rejected applications—not only the candidates the system recommends; 

remove information that is irrelevant to performance and may act as a proxy for age or another protected attribute; 

give candidates a practical way to request human review or reasonable adjustment; and 

hold internal decision-makers accountable for outcomes rather than treating the technology provider as responsible for fairness. 

Use AI to strengthen judgement—not replace it 

The answer is not to reject AI. Used well, it can help employers manage volume, improve consistency and create more time for meaningful human assessment. 

But it must remain a tool in service of sound judgement. It should help recruiters see talent more clearly, not narrow the definition of talent to whatever the algorithm finds easiest to recognise. 

The Today Show story is powerful because the woman did not become a better candidate when she removed the dates from her resume. She simply became visible. 

Before celebrating the efficiency of your recruitment technology, ask a harder question: who is it preventing you from ever meeting? 

Sources and further reading 

• Australian Human Rights Commission, “AI and recruitment compliance checklist” 
• Dr Natalie Sheard, University of Melbourne, “Discrimination by recruitment algorithms is a real problem” 
• Nine, “Expert warns women are being left out in the cold as more employers use AI for recruiting”