Skip to Content
FREE TOOL · INSTANT ESTIMATE

How much is AI cheating distorting your shortlist?

When candidates use AI to inflate their assessment scores, they don't just cheat — they crowd out your genuinely strong applicants. This calculator estimates how many shortlist places are being taken by AI-assisted candidates under your own assumptions.

  • clock icon Instant result
  • document icon 4 inputs
  • user tick icon No email required
Ben Schwencke, Chief Psychologist
Built by Ben Schwencke Chief Psychologist, Test Partnership

Estimate your shortlist exposure

Total candidates completing the assessment.
This creates the selection ratio. For example, 100 selected from 1,000 tested = 10%.
Estimated proportion of candidates who use AI assistance.
Average percentile rank achieved by AI-assisted candidates. 50th percentile = non-cheater average; 84th percentile is roughly one standard deviation above average.

Estimated shortlist impact

- Share of shortlist who cheated
- Cheater overrepresentation
- Expected cheaters in shortlist

Risk level

Adjust the inputs to estimate shortlist distortion.

Shortlist composition
Cheaters shortlisted
-
Non-cheaters shortlisted
-
hero bar overlay
Why this matters

A small cheating rate can take a large share of your shortlist.

AI-assisted candidates don't score randomly — they cluster near the top. Because shortlisting selects from the top down, even a modest proportion of cheaters can be heavily overrepresented among the people you actually invite forward. The genuinely strong candidates they displace never make your shortlist.

This calculator makes that effect concrete. Set your own volumes and assumptions, and it estimates the share of shortlist places taken by AI-assisted candidates — and how overrepresented they are relative to the wider applicant pool.

How to read it

Four inputs, three numbers that matter.

1

Set your volumes

Enter how many candidates you test and how many you shortlist. Together these set your selection ratio.

2

Estimate the cheating

Set the proportion of candidates using AI, and how well they score on average as a percentile.

3

Read the share

See the estimated share of shortlist places taken by cheaters, and how overrepresented they are versus the pool.

4

Gauge the risk

An honest risk band tells you whether AI is a minor nuisance or quietly invalidating this stage of your process.

About the model

The calculator models scores as a mixture of two normal distributions: non-cheaters centred on the population average, and AI-assisted candidates shifted up to the percentile you set. It finds the score threshold that produces your selection ratio, then works out how many of each group fall above it. It's a directional estimate to support a conversation about AI risk — not a precise measurement of any individual process. The maths runs entirely in your browser; nothing you enter is stored or sent anywhere.