The cheapest way to practice a police promotion board used to be reading your STARR answers back to yourself in the car. The most expensive was paying a firm a few hundred pounds for a single human mock board. AI has quietly opened a third way to practice for a police promotion board, and it beats both, but only if you use the right kind of AI.
Here’s what good AI practice looks like, and what to avoid.
Reading your answers isn’t the same as sitting the board
Most preparation happens in silence. You write an answer, tidy it, read it back. That’s useful, and if you’re still getting the structure right our guide on how to write a STARR answer is the place to start.
But the board isn’t silent. You say your answer out loud, to people, under time pressure, and then they probe it. A written answer that looks strong on the page often falls apart the moment you’ve got to deliver it and defend it. The probe is where a learned script breaks, because the panel asks about the part you skipped. Delivery is the thing most strong officers never actually rehearse.
Generic AI won’t get you there
Pasting an answer into ChatGPT feels like practice, but it isn’t. It marks text, not delivery. It can’t hear you speak, it can’t follow up on what you said, and it doesn’t know the CVF 2024 descriptors your panel is scoring you against. It doesn’t know your force’s HMICFRS inspection picture either, which the questions on a real board can turn on.
It’s also kind when a panel won’t be. We went into the detail of why ChatGPT won’t get you promoted in a separate guide. The short version is that it has no rank ceiling and no reason to tell you the truth.
What good AI practice actually does
Real practice has to do what a panel does. It should ask the question out loud, listen to the answer you actually gave, probe the weak part of it, hold you to board timings, and mark you against the Competency and Values Framework at the rank you’re going for.
Then it should tell you where the marks went. A useful debrief gives every answer a mark out of five, names the structure you used, counts how often you said “we” when the panel needed “I”, and flags the filler words you didn’t notice. That’s feedback you can act on, not a pat on the head.
Rank calibration matters more than officers expect. An answer pitched below your target rank scores poorly however polished it sounds, which is why the feedback has to be tied to the rank. If you’re not sure how far the bar moves, the guide on what panels look for at inspector level shows it clearly. State6’s AI Mock Board does all of this: a voiced panel that asks aloud, probes, times you and marks every answer out of five with a written debrief.
Practice on your own time, not someone else’s diary
A human mock board means booking a slot, usually for one sitting at a few hundred pounds. If you work shifts that’s hard to arrange and harder to repeat.
AI practice runs in your browser whenever it suits you, whether that’s after a late turn or in a quiet half hour at home. The point is repetition. You don’t fix your delivery in one go, and one expensive sitting never gave you the chance to.
The goal is simple
You want the first time you answer a board question out loud, under pressure, to be a rehearsal and not the real thing. That’s the whole case for practising with AI, as long as the AI behaves like a panel rather than a cheerleader. If you want to see what behaving like a panel means under the hood, we break down how we do what we do.
For the full picture of the process, from how the NPPF works to what panels score at each rank, see the complete guide to UK police promotion boards.