Your first sift cuts the most and measures the least: Measure the person, not the AI-written CV

You cut more people at the first sift than at every other stage of your hiring process combined. It’s also the stage where you know the least about any of them. That cut is usually made on a CV, and in 2026 that CV is often written by AI, read by AI and ranked by AI.

In this on-demand session, Chief Commercial Officer Hannah Mullaney is joined by Chief Science Officer Rab MacIver. They look at what a CV sift can and cannot tell you about a candidate, why AI screening is making the problem harder to see, and what the top of your funnel should measure instead.

A study of 2,200 CVs published in June 2026 found the machines doing the reading prefer the CVs the machines wrote, by up to 82%. And when Duke University researchers analyzed 200,000 CVs from a live hiring platform, around 1 in 100 carried hidden instructions telling the software to mark the candidate as qualified.

Rab explains why the individual parts of a CV, like years of experience and education, predict job performance only slightly better than chance, and why handing them to AI does not make them any better. The session closes with the fix we see: a short measure of the person that sits inside your ATS, and is cheap, valid, reliable and effective at scale.

What you’ll learn

• What the top of your funnel actually measures, and what a CV can and cannot predict
• How AI-written CVs, hidden keywords and hidden instructions game screening software
• Where bias creeps into AI screening, and why that matters for defensible hiring
• What a top-of-funnel measure needs: fast for candidates, simple for recruiters, fair and role specific
• What changes when you size up the person instead of the paperwork