If you are using ChatGPT to prepare for your police promotion board, let’s be honest with each other.
You copied a STARR answer in, hit send and got back something that sounded polished, confident and vaguely impressive. ChatGPT told you it was strong. Maybe suggested a tweak or two. You felt slightly better about your preparation and closed the laptop.
We get it. We’ve done the same thing ourselves.
Here’s the problem though. ChatGPT was being kind. And promotion panels won’t be.
ChatGPT Is a Brilliant Tool. Just Not for This.
We want to be clear before we go any further. ChatGPT is genuinely impressive and we are not here to knock it. It can draft emails, summarise documents and explain complex ideas in seconds. If you asked it to help you write a leaving speech for your skipper it would probably do a decent job.
But preparing for a UK police promotion board is not a general task. It is a very specific one. And the things that make the difference between passing and failing are things ChatGPT simply does not know.
What ChatGPT Doesn’t Know
It doesn’t know the CVF 2024 descriptors
Your panel is not listening to your answer and thinking “that sounds good.” They are sitting with the CVF 2024 framework in front of them scoring you against specific positive and negative indicators for each competency.
ChatGPT has a general awareness of competency frameworks. It does not know the precise descriptors your panel is using on the day. So when it tells you your answer covers “We Take Ownership” well it has no way of knowing which specific indicators you hit and which you missed.
State6 scores every answer descriptor by descriptor. Not a general impression. The exact framework your panel uses.
It doesn’t know your force
Your panel knows your force’s inspection picture. They know what HMICFRS said. They know the causes of concern and where the force has been told to improve. A candidate who weaves that context naturally into their answers shows the kind of strategic awareness that gets you promoted.
ChatGPT knows nothing about your force’s most recent PEEL inspection. It cannot tell you that your force has a Requires Improvement grade for investigations and that your answer should reflect that. It cannot connect your evidence to what your chief officers are actually focused on right now.
State6 has all 43 forces loaded. Your force’s real inspection picture is built into every review and every coaching session.
It doesn’t know what rank you are going for
A sergeant answer and a chief inspector answer to the same question should sound completely different. The level of strategic thinking, the language, the scope of the decisions you describe — all of it needs to be calibrated to the rank you are applying for.
ChatGPT will tidy up your answer. It will not tell you that your answer sounds like a constable when you are going for inspector. It has no rank ceiling to apply and no calibration built in.
State6 knows your target rank. Every piece of feedback is calibrated to it.
It won’t catch your collective language
“We did this.” “Our team decided.” “We worked together to resolve it.”
If you have ever written a STARR answer you have probably done this without even noticing. Most officers do. Panels hear collective language and immediately start wondering what you personally did. ChatGPT will read straight past it. Sometimes it will even add more of it trying to make your answer sound collaborative. We dig into why “we” costs you marks in a separate guide.
State6 catches collective language and flags it every time. Because panels do.
It treats every question as a behavioural question
Promotion boards ask two types of question. Behavioural questions draw on past experience and STARR is the right structure. Forward facing questions ask how you’d handle something in a new role and need a different structure entirely. ChatGPT won’t make that distinction. It will prompt you to give an example when the panel wants a position.
It tells you what you want to hear
This is the big one.
ChatGPT is designed to be helpful. Kind, even. It is not designed to tell a serving police officer that their answer is weak, that they pitched it at the wrong rank or that they failed to evidence a single positive indicator for the competency they were asked about.
That kind of honest feedback is uncomfortable. But it is exactly what you need before the board. Not after it.
If you are not sure how to structure your answers in the first place, start with our guide on how to write a STARR answer for a police promotion board.
What It Actually Looks Like Side by Side
An inspector going for chief inspector submits a STARR answer on We Are Emotionally Aware.
ChatGPT response:
“This is a strong answer that demonstrates good emotional awareness and leadership. You clearly show empathy towards your team and handle the situation with sensitivity. Consider adding a brief mention of how you monitored the outcome over time.”
Sounds reasonable. Feels helpful. Scores poorly.
State6 response:
“We Are Emotionally Aware — scored 2 out of 5. Leading With Integrity descriptor: not evidenced. No reference to your own emotional response or how you managed it in the moment. Collective language detected in Action section — three instances of ‘we’ where the panel needs ‘I’. Rank calibration: this answer reads at sergeant level. A chief inspector response needs to demonstrate awareness of the wider organisational impact, not just the immediate team dynamic. Suggested revision focus: your personal emotional response, a specific moment where you had to regulate under pressure, and the force context around workforce wellbeing.”
Same answer. Completely different feedback. One of those gets you promoted. The other makes you feel good about an answer that would score a two on the day.
The Thing No AI Can Replicate
State6 was not built by a product team who read about police promotion boards online.
It was built by a serving Chief Inspector and Inspector with 38 combined years of frontline experience across four forces. Every tool, every scoring framework, every piece of coaching logic has been personally designed and checked by people who have sat on promotion panels and watched candidates pass and fail in real time.
We know what panels are looking for because we have been the panel. We know what a two out of five sounds like because we have scored them. We know the difference between an answer that impresses and one that just fills the time because we have heard both more times than we can count.
ChatGPT does not have that. It cannot have that. It is a generalist tool doing its best with a specialist task and there is no criticism in that. It is just the truth.
You already know ChatGPT is not built for this. That is probably why you are reading this article.
If you want to rehearse out loud rather than just check your text, this is the case for practising a police promotion board with AI. And if you want to see how a built scoring system differs from a chatbot under the hood, we explain how we do what we do.
If you are also preparing a board presentation, the guides on why presentations fail before the first slide and the STANCE framework cover the structure side of it.
State6 is.
For everything on the promotion process itself — how the NPPF works, what the CVF requires and how to structure every question type — see the complete guide to UK police promotion boards.