AI has become standard in hiring. As of 2025, 51% of organizations now use AI in recruitment, according to SHRM’s Talent Trends report. Imagine how that number has shifted in just a single year. The efficiency gains behind the trend are real. But candidates aren’t buying it wholesale. A Greenhouse study found only 8% of job seekers believe AI screening makes hiring fairer. Half have less trust in the application process than they did a year ago.
Where automation breaks down
The core failure point is keyword matching. Even sophisticated systems can filter out strong candidates simply because their resume doesn’t use the exact phrasing an algorithm was trained to look for. A candidate can have every skill a role requires. They can still get auto-rejected because their experience wasn’t described in the language the system was trained to recognize. Unless someone catches the error, that candidate is gone before a recruiter ever sees the resume.
Many recruiters rely on Boolean searches to comb internal systems. AI models can run these same searches, often faster. But job titles and responsibilities shift across companies and industries. Candidates pick up niche skills along the way that a search string was never built to catch. A search built around exact terms misses the person who’s done the work under a different title. It also misses the person who developed exactly the right capability somewhere the query didn’t think to look.
Where a person in the loop matters
Full automation makes filtering decisions final. A recruiter makes them a conversation. AI provides the insight. Human recruiters add judgment, empathy, and compliance oversight, none of which an AI model currently replicates.
The stakes go beyond individual missed hires. Recent studies, including large-scale research out of Stanford, are showing that certain AI hiring models can introduce racial bias into candidate screening. This bias can compound across employers who rely on the same vendor. Human recruiters aren’t immune to bias either, but that bias isn’t hardcoded into a system applying the same flawed logic to millions of applications at once. A person can still ask questions, weigh context, and catch a mistake before it becomes a pattern. That’s something a fixed algorithm structurally can’t do on its own.
This is where RPO earns its value over a fully automated platform. The technology should still do what it’s good at: handling volume, surfacing patterns, and speeding up the parts of the process that don’t need judgment. But the decisions that actually determine hire quality still benefit from a person applying context an algorithm doesn’t have.
What this means for employers
If your current hiring process leans fully automated, it’s worth asking a direct question. How many strong candidates are being filtered out before a human ever sees them? And just as important, how much trust is eroding on the other side of that process? Efficiency numbers look good on paper. Lost trust and lost candidates are the cost you don’t see.
At Lucas James Talent Partners, every candidate is reviewed by a real person who can ask questions, notice context, and make a judgment call an algorithm can’t. That’s not a limitation. It’s the reason our clients get better hires.