The client needed to hire a junior analyst and understand honestly what the person could do. The difficulty is that answers to an ordinary test task can now be "boosted" with an AI model in a minute, and the test stops measuring anything. The question was direct: how to assess a junior so the result can be trusted.
Why a classic test for a junior no longer works
A test task used to check whether a candidate could reach the right answer. The right answer is now available to anyone in a couple of seconds, and the task measures the ability to use an AI model, not thinking. For a junior this is especially painful: at the start what matters is seeing how a person thinks, not what text they can generate.
How we built a screen for this risk
We made a calibrated junior screen: questions with clear markers — what counts as a green signal, what as a red one — and video-recorded answers, to see the line of reasoning rather than only the result. We did not hide the "answers via AI" risk; we named it and built it into the question design. We checked systematic thinking and how a person explains their decisions, not memorised phrasing.
What happened
The client got results comparable across candidates and protected against gaming. The hiring decision became well-grounded: what was visible was not "who wrote it more nicely", but who thinks more clearly and is more honest about what they do not yet know. For a junior the second matters more than the first.
What follows from this
A junior is worth checking for systematic thinking and communication through tasks with clear markers, not for the ability to produce the right answer. In a world where the answer is available to everyone, a candidate's value lies in how they reason and where they honestly admit the limits of their knowledge.
We will build a screen for your role — protected against gaming, with clear markers. Email [email protected] or message @Vasiliadi on Telegram — we will look at your situation.
Email Anastasia →