A Chief Human Resources Officer I know told me something recently that has stayed with me.

“We have more data on candidates than ever, and somehow I trust my hiring decisions less than I did 10 years ago,” she said.

She admitted it quietly, like an admission she was not supposed to make.

In conversations with executives across industries, I hear the same concern, and it is not a lack of experience or information that is driving it. If anything, leaders now face the opposite problem – there are more resumes, more credentials, more assessments and more AI-powered platforms promising to optimize every stage of hiring, yet confidence in hiring decisions, especially for early career roles, has quietly eroded.

AI did not create this problem, it revealed it.

Most hiring systems were built to optimize throughput, not truth. Applicant tracking systems filter for keywords, interviews reward fluency and confidence, and assessments test narrow fragments of skill in artificial settings, so each step produces more data but very little clarity about how someone will actually perform once the job begins.

The consequences rarely appear at the moment a hiring decision is made; they show up months later, when a hire who looked exceptional on paper struggles when conditions change, the team quietly compensates and the manager spends hours course correcting instead of building the business.

Go beyond surface signals

Research estimates the average cost of a bad hire at about US$7,000 and as high as 30 percent of first-year earnings once lost productivity, recruiting costs and training are considered, which for a US$100,000 role can mean US$30,000 in damage before accounting for delayed projects, morale erosion or missed opportunities. These are not small inefficiencies; they are balance sheet impacts disguised as HR problems.

It is important to state one thing clearly: most candidates are not acting maliciously. When people polish resumes with AI or optimize their profiles for algorithms, they are responding rationally to the incentives of the system. The issue is structural; hiring infrastructure was built for a world where surface signals such as credentials, polished narratives and confident interviews were reasonably reliable indicators of capability and that correlation is weakening.

A recent Checkr survey of 3,000 hiring managers found that 59 percent suspect candidates are using AI tools to misrepresent themselves in resumes, interviews or written assessments, and Gartner projects that by 2028 one in four candidate profiles could be entirely fabricated. When nearly six in 10 hiring managers question the authenticity of what they are seeing, the hiring signal itself begins to break down.

There is also an equity dimension to this shift that I find difficult to ignore, both as a practitioner and as a parent. Systems that reward polished credentials and carefully packaged narratives tend to advantage those who understand how to play the signaling game, while candidates with real ability but fewer resources or connections may struggle to present themselves in ways that appear competitive, which means the people most likely to be screened out are often not the least capable ones.

The instinctive response to this uncertainty is to add more screening layers, more assessments, more interviews, more AI filters, but layering technology onto a fragile signal rarely restores trust; what restores confidence is changing what counts as evidence.

Organizations that are regaining trust in their hiring decisions are making a shift that is simple in concept but requires real commitment in practice: instead of evaluating candidates primarily on what they claim they can do, they create opportunities to observe how candidates actually work.

This often means structured, role-relevant tasks that allow candidates to demonstrate how they frame problems, interpret information and make decisions under realistic constraints, with the goal not being to create a perfect simulation of work but to provide enough structure that leaders can compare candidates based on observable thinking and judgment rather than the impression a polished interview leaves behind.

Evidence-based decisions

When hiring conversations are grounded in real evidence, something worth noting tends to happen; debate becomes shorter, decisions become clearer and leaders are no longer arguing about impressions but discussing what they have actually seen.

Better evidence does not replace human judgment, it strengthens it, because leaders still make the decision, but they make it with something more reliable to stand on.

Every CEO faces the same tension between moving fast and deciding well, between scaling efficiently and preserving trust. Those tradeoffs feel unavoidable because most hiring systems force them to be.

Organizations that anchor hiring decisions in observable work, however, often discover they can move faster and decide better simultaneously, spending less time debating assumptions and far less time unwinding mistakes that better evidence could have prevented.

In an AI-saturated world, the competitive advantage will not belong to organizations with the most candidate data or the most sophisticated screening tools; it will belong to organizations that know what evidence they can trust, and that distinction is worth building toward deliberately rather than waiting until the cost of guessing makes it impossible to ignore.

Opinions expressed by The CEO Magazine contributors are their own.