AI has made it easier for candidates to look qualified. The harder question is whether the signals companies use to separate them still work.
At least 1% of the resumes submitted to one hiring platform contained hidden instructions written to manipulate the AI reading them. Researchers from Duke, Arizona State, Berkeley and UNC found them across roughly 200,000 real resumes, and attempts rose sevenfold between July 2024 and November 2025.
The problem extends past resumes. In a Gartner survey of 3,000 candidates, 6% admitted to interview fraud, either posing as someone else or having someone else pose as them. Gartner predicts that by 2028, one in four candidate profiles worldwide will be fake.
Those numbers deserve attention, but set fraud aside for a moment. There is another change affecting nearly every search: almost everybody can now present well.
The filter still works. It just stopped separating.
Think of a filter built to separate large stones from small ones. If every stone becomes the same size, the filter is not broken. It is operating exactly as designed. It has simply stopped separating anything useful.
That is increasingly the problem with traditional candidate screening. A polished resume used to carry information. So did a carefully written summary or a strong first conversation.
Those signals are now inexpensive to produce. AI can rewrite a resume against a job description, sharpen an executive summary, structure a career history and prepare a candidate for the questions likely to come next.
Gartner found that 39% of surveyed candidates used AI during the application process, and 54% of those used it to generate resume or CV text.
That does not make those candidates dishonest. It means presentation is becoming a weaker proxy for capability. The resume is clean, the summary is sharp and the first screen goes fine.
The question is what any of it still proves.
Looking qualified is not the same as being qualified
Consider a VP candidate with the right title, the right industry experience and a resume full of growth claims. The document may tell you the person increased revenue, built a team or turned around an operation. It does not tell you enough about the conditions under which they did it.
A VP at a $40 million company is not automatically comparable to a VP at a $4 billion company. Growing a business from $20 million to $40 million is not the same assignment as growing one from $400 million to $800 million.
Those differences do not fit into keywords. They require context.
That changes the screening question. Instead of asking, “Does this candidate’s background match the role?” the better question is, “Does this person’s experience match the problem this company needs solved?”
The cost appears after the hire
That distinction matters most when the seat matters. If the wrong person reaches a first interview, the cost is small. If the wrong person starts the job and spends six months in a role tied directly to a value creation plan, the calculation changes.
A commercial initiative stalls. An operational improvement slips. Leadership spends months believing a problem has been delegated when it has not been solved.
The recruiting fee is rarely the biggest cost in that situation. The bigger cost is the time the company thought it was making progress.
For a critical hire, getting someone into the seat quickly is only half the equation. The other half is having enough evidence to believe that person can deliver what the seat exists to accomplish.
Screening has to move closer to the work
If presentation is easier to manufacture, evaluation has to move closer to evidence. That does not mean adding more interviews. It means getting closer to what the person actually did.
For a sales leader, do not stop at “grew revenue 30%.” Understand the market they inherited, what they changed, which part of the result was theirs and what happened when the plan missed.
For an operations leader, get underneath the improvement percentages. What was broken, what did they personally own and what changed because of their decisions?
For a finance leader, go past the transaction or forecast listed on the resume to the decisions behind it and their actual responsibility for the outcome.
Four questions tend to reveal more than another hour of polished interviewing: What was the problem? What did you own? What changed because of you? Who can verify it?
The goal is not to make candidates perform. It is to replace presentation with evidence.
AI and Human Intelligence
At TAG, we think about the process in two parts: Artificial Intelligence and Human Intelligence. AI and HI.
AI is genuinely useful. It handles research, organizes information, identifies patterns and removes administrative work that used to consume the early stages of a search.
Human Intelligence carries a different part of the decision. It is knowing why the same title means two different things at two companies. It is noticing when the numbers on a resume do not line up with the scope of the role. It is knowing which follow-up question to ask because something in the story does not quite fit.
Most of that judgment comes from understanding the business well enough to know which evidence matters. That is a different exercise inside a portfolio company hiring against a specific operating objective than inside a company simply filling a vacancy.
Technology makes the process faster. Human Intelligence determines whether the output means anything.
AI did not break hiring
It exposed how heavily hiring depended on signals that were easier to manufacture than companies realized.
The resume still matters. Interviews still matter. AI has a role. But none of them should be confused with proof.
When almost every candidate arrives polished, the advantage belongs to the company that knows what to look for beneath the polish.
Move quickly, but know what the signal is telling you. Speed without signal is just faster uncertainty.