Why Hiring Feels Harder Than Ever, Even With AI

As artificial intelligence floods applicant tracking systems with polished applications, companies are confronting a deeper crisis: many still have not defined what real competence looks like on the job.

A mid-sized technology company that posts a product management opening can receive 800 to 1,500 applications within 72 hours. On paper, many look perfect. Résumés mirror the job description, portfolios present polished metrics on revenue growth and “cross-functional alignment,” and the keywords applicant tracking systems favor are all there.

But when it is time to make the hire, many teams face the same problem: despite the volume and the polish, they still cannot tell who will actually do the job well.

“We are in an era where nearly everyone looks qualified on paper, and the résumé is no longer a reliable signal,” says Elina Konstantinidou, founder and CEO of Noscis.ai, a behavioral science consultancy that helps growing companies build organizational systems. “Companies are getting more applications than ever, spending more on automated screening than ever, and learning less about real ability each time they hire.”

Across corporate America, the response has been predictable. Companies have added more screening software, more automated résumé parsing, and more online assessments, including recorded interviews scored by algorithms that analyze speech and facial cues. Investors, meanwhile, have poured millions into AI platforms that promise to fix hiring by filtering candidates faster and with greater “accuracy.”

Behavioral scientists say this is aimed at the wrong target. The core problem is not the time it takes to review thousands of résumés. The problem is that many organizations start hiring before they have defined what success in the role actually looks like. Without clear standards, even the most advanced tools can only process uncertainty at scale.

Research suggests the problem is widespread. A 2024 Gallup workplace study found that fewer than half of employees strongly agree they know what is expected of them at work. When expectations are unclear inside an organization, hiring becomes unclear too. Job descriptions are often stitched together from competitors’ postings or generic HR templates. Interview panels frequently enter the process without shared criteria, then end up judging candidates on confidence, rapport, or familiar networks rather than demonstrated ability.

“Organizations treat recruitment as an administrative exercise in filling an empty desk rather than an architectural problem,” Konstantinidou said during a session at the Women in Tech Global Conference earlier this year, where she spoke on: Hiring is the hardest thing founders get wrong. AI is making it worse. “If you do not know which specific behaviors produce superior outcomes in your specific environment, every candidate screening tool you buy simply automates your own confusion at higher volume.”

For decades, the standard fix has been competency libraries sold by consulting and human capital firms. These packaged lists include terms such as “strategic agility,” “collaboration,” and “results orientation.” Companies pick a few, paste them into performance reviews, and ask managers to rate people accordingly.

In practice, these labels often fail to predict who will succeed.

Take two roles that might both be labeled “commercial acumen.” A business development leader at a 40-person software company has to live with uncertainty, build the sales motion from scratch, and figure out why prospects hesitate. An enterprise sales executive at a public company, by contrast, may work within strict compliance rules, navigate long approval processes, and coordinate closely with legal and procurement. Using the same label for both roles obscures what actually matters: the day-to-day behaviors that determine whether a hire succeeds or is gone within months.

To close that gap, Konstantinidou developed what she calls the Noscis Behavioral Competency Framework, a method built around three phases: uncover, define, and measure.

Instead of selecting traits from a generic list, the process begins with structured discovery inside the organization to identify the behaviors that separate high performers from average ones in that specific context and for that specific role. Those insights are then turned into clear, observable benchmarks. Finally, the benchmarks are built into interview rubrics and scorecards that aim to produce consistent evidence, not gut feeling.

The problem extends beyond employers. It is also shaped by how much of the recruitment industry is paid.

In the US, many search firms are paid through placement fees, often 20% to 33% of a candidate’s first-year base salary. That model rewards speed. Once an offer is accepted and the guarantee period expires, the agency usually has little accountability for whether the hire performs well, fits the team, or leaves within the first year.

“When financial incentives reward seat-filling, the agency’s goals and the client’s goals inevitably diverge,” Konstantinidou says. “The recruiter is incentivized to find someone acceptable enough to clear the interview panel quickly. The employer needs someone who will drive business results over the next three years. If you want sustainable hiring, you have to build an internal system that outlasts any single recruitment campaign, rather than renting outside recruiters on a transaction-by-transaction basis.”

Importantly, this does not mean companies should abandon artificial intelligence. It means they should use it differently. “AI should come in after you’ve decided what good looks like, not before,” Konstantinidou said. “Used that way, it makes your standard sharper and more consistent across interviewers. Used the other way, it just makes your guess faster.” AI tools can then support structure and execution by synthesizing interview notes, flagging inconsistencies in scorecards, and helping teams draft rubrics quickly, while reducing administrative work and making hiring easier to manage.

But the final evaluation still depends on trained human judgment. Determining whether someone shows sound judgment under pressure, takes feedback well, or handles conflict constructively requires careful observation and structured questioning. Software can support that work, but it cannot replace it.

For leaders struggling with growing recruitment costs and persistent turnover, the message is becoming hard to dismiss. The companies that succeed in the current hiring landscape will not be those with the most elaborate filtering systems. They will be the organizations willing to do the harder work first: clearly defining what competent execution looks like, then hiring against that standard with discipline and consistency.