Skip to content
Hiring

Recruiters find candidates faster with AI, trust them less

A grey-haired recruiter in a rust sweater checks a printed shortlist with a pen beside a laptop showing candidate profiles.

What the survey found

AI has made sourcing faster for most recruiters, but it has not made them surer of the people they find. That is the main finding of the Recruiting in the AI era 2026 report from TestGorilla, an assessment platform, which surveyed 1,428 US talent professionals in August 2026 and 483 in the UK in September and October.

  • 72% of US respondents said AI increased the speed of finding candidates over the last 12 months.
  • Of those who got faster, about a third reported no gain in certainty that they had found the right person.
  • 18% of US respondents said AI made their confidence in a match fall; in the UK the figure was 24%.
  • 67% of US respondents spend an hour or more checking a single candidate before they trust that person can do the job, and 29% spend two hours or more.
  • 57% of US respondents named verifying whether resume skills are real as a top-three sourcing challenge, the most cited answer.

HR Dive covered the survey in its coverage of the survey. Only 48% of US respondents said AI made them both faster and surer.

What the numbers do not prove

This is a self-reported survey, not a measure of placements. It tells you what recruiters feel about speed and confidence, not whether AI-sourced candidates were hired, kept, or performed worse. Nothing in the report links a sourcing tool to a failed placement.

The sample also needs a careful read before an agency applies it. About a quarter of respondents were external recruiters; the rest worked in talent acquisition or HR. The study comes from a vendor that sells skills tests, so its framing favours objective checks. None of that makes the figures wrong, but it limits how far they travel.

Why the manual check still takes an hour

Sourcing tools widen the pool, and most of what they return is still what the candidate wrote about themselves. A profile that lists a skill says nothing about whether anyone tested it. So the recruiter phones, reads, and cross-checks by hand, and the time saved at the top of the funnel moves to the middle.

Volume makes this heavier. Ashby's 2026 recruiter productivity analysis found that applications per hire tripled from 2021 to 2024 and stayed above 300 through 2025. The check also matters for trust in the person behind the profile, a risk covered in our piece on candidate fraud in remote hiring.

One change: show the client what you checked

For an agency, the hour of checking is already being spent. The change is to make it visible. A shortlist that looks like a list of profiles hides the work that separates it from what the client could pull from a database.

A simple way to do that is to mark every claim on each shortlisted candidate with how it was confirmed:

  1. Checked: tested, or confirmed with a reference, a work sample, or a document you saw.
  2. Stated: the candidate said it in an interview and it held up under questioning.
  3. Self-reported: it appears on the CV or profile and nobody has confirmed it yet.

The client then sees where the risk sits and can decide which gaps to close before the interview. It also gives the agency a fair answer when a client asks why it should pay for candidates a tool could find. Some of those checks may come from automated screening, and our article on candidates taking AI screening interviews after hours looks at how that step fits a candidate's day.

What to watch

The report says 92% of US respondents would be willing to run some form of objective check on a sourced candidate before an interview. If clients start to expect that, a shortlist without a verification column will look unfinished. Agencies that already label what they checked will not need to change much.

Related articles

Comments

No comments yet

Be the first — ask a question or share your experience with this article.

Write a comment