AI skills tests for job applications

More UK employers now include some kind of AI task in their hiring process. It might be a short practical exercise, a scenario question at interview, or a request to show how you would use a tool such as Copilot or ChatGPT. This guide explains the common formats, how to prepare in a week, and how to talk about your result without overstating it.

What "AI skills tests" usually look like

Formats vary a lot between employers and roles, but most fall into a few types:

A practical task: you are given a scenario and asked to write the prompt you would use, or to produce a piece of work with an AI tool and explain your steps.
A review task: you are shown AI-generated output and asked to spot errors, risks or missing information.
A scenario interview question: "How would you use AI to speed up this process?" or "When would you not use AI here?"
A take-home exercise: complete a short piece of work, sometimes with permission to use AI, and say how you used it.
Online assessments from third parties: these are often used for screening, with a score or pass mark.

What assessors are looking for

Across these formats, assessors tend to look for the same qualities. They want a clear brief: does your prompt say who the output is for, what is needed and what to avoid? They want judgement: do you check facts and figures, and notice when AI is the wrong tool? They want responsibility: do you avoid pasting personal or confidential data into tools that are not approved? And they want communication: can you explain what you did simply?

Notice that none of these is about knowing technical jargon. A candidate who writes a careful, specific prompt and explains how they would check the result usually beats one who uses impressive terms.

A one-week preparation plan

Day 1: read the job advert and note the tasks the role involves. These are the scenarios you are most likely to be given.
Day 2: pick three of those tasks and practise each with an AI tool, using made-up data. Write your prompt, review the output, and improve it twice.
Day 3: practise spotting problems. Ask a tool for an answer on a topic you know well and list every mistake or vague point you can find.
Day 4: prepare a short answer to "When would you not use AI?" (for example sensitive personal data, legal or medical judgements, or anything you cannot verify).
Day 5: take a timed practice assessment so the format feels familiar.
Day 6: write one half-page case study of a real task where AI helped you, including how you checked the result.
Day 7: rest, and reread your notes.

How to answer a practical prompting task

When you are asked to "write the prompt you would use", work through this checklist before you submit:

State the goal and the audience in the first line.
Give the context the tool needs, and say what it must not assume.
Handle data carefully. If the scenario involves customer or personal data, say that you would remove or anonymise identifying details first, or use only an approved tool. Assessors often look for this.
Specify the format: length, structure and tone.
Build in checks: ask the tool to flag uncertainty and list assumptions.
Add one line on how you would verify the output before using it.

Example structure:
"You are helping a UK retail manager prepare a summary for head office. Using only the anonymised sales figures below (customer names removed), identify the three biggest changes since last month and suggest one possible reason for each, clearly marked as hypotheses. Format: five bullet points, plain English. Flag any figures that look inconsistent. Data: [anonymised table]"

Common mistakes

Writing a one-line prompt for a complex task.
Ignoring the personal data in the scenario.
Accepting the AI output as correct without checking.
Overclaiming: saying AI "always" saves hours, or calling yourself an expert.
Forgetting the human: not saying who reviews the output before it is used.

Presenting your result honestly

If you have taken an assessment, mention it accurately. Say who provides it, what it measured and what the result was. Do not describe a private skills credential as a regulated qualification or accreditation; employers notice, and honesty builds trust. A link the employer can open themselves is more persuasive than a claim in your covering letter.

The free PromptRank assessment from Busigurus is a practical prompting task marked by a fixed rubric, not by AI, so the same answer always gets the same score. If you score 60 or more you can optionally get a £29.99 credential with a public verify link to include in applications. It is a proprietary skills credential, not a regulated qualification.

If the employer does not test AI skills

You can still show them. Add one achievement bullet to your CV that links AI use to an outcome and mentions how you checked the work. Prepare one short example for interview. Many employers are interested even when the advert does not mention AI.