An AI skills test measures how well people use AI tools in realistic work: writing good prompts, checking what the AI produces, using judgement about when to rely on it, and fitting it into a workflow. This guide explains why these skills matter, what AICheck measures, a generic example, and how employers can use the results.

Why AI skills matter

AI assistants are now part of everyday office work, from drafting emails to summarising documents and analysing data. Used well, they save time. Used carelessly, they produce confident errors that end up in reports, client letters and decisions.

A field experiment with consultants by Dell'Acqua and colleagues showed both sides: AI improved speed and quality on tasks it was suited to, but people who relied on it for a task outside its capabilities were more likely to get the answer wrong [1]. The skill that matters is not just using AI, but knowing when and how to trust it.

What AICheck measures

AICheck (AI Skills Test) places candidates in realistic AI-assisted work scenarios. They write prompts, put workflows in order, fact-check AI output and make judgement calls, working with a scripted in-test Assistant so every candidate sees the same responses. It is built on the UK AI Skills for Business Competency Framework and reports score bands for four skill areas:

• Prompting

• Verification

• Judgement

• Workflow

The Core level takes 35 minutes and is scored automatically. AICheck is delivered on iTestHub, Viewpoint's sister testing platform.

A worked verification example

The following is a generic illustration written for this article, not a task from AICheck.

You ask an AI assistant to summarise a supplier contract. The summary says the notice period is 30 days, payment terms are 60 days, and the contract renews automatically each year. The task is to decide which points you must check against the original before sending the summary to your manager.

All three are facts that matter and that an AI could get wrong, so all three should be checked against the contract itself. A candidate who checks only the figure that "looks odd", or none at all because the summary reads well, shows a gap in verification, which is one of the most important AI skills at work.

Using an AI skills test well

For employers, AICheck suits office and admin roles, analysts and researchers, and managers: anyone expected to use AI tools responsibly. Use the skill-area bands to focus training. Someone strong at prompting but weaker at verification, for example, may benefit most from guidance on checking AI output.

For candidates, try the example questions first. Treat the in-test Assistant as you would a capable but fallible colleague: give it clear instructions, check its work, and decide for yourself.

See the AICheck page, our PyCheck (Python Coding Test), or all our work skills tests.

Frequently asked questions

How long does AICheck take?

The Core level takes 35 minutes.

Does it use a real AI model?

It uses a scripted in-test Assistant, so every candidate sees the same responses and scoring is fair.

What framework is it based on?

The UK AI Skills for Business Competency Framework.

What does the report show?

Score bands for Prompting, Verification, Judgement and Workflow.

References

1. Dell'Acqua, F., McFowland, E., Mollick, E. R., Lifshitz-Assaf, H., Kellogg, K., Rajendran, S., Krayer, L., Candelon, F. & Lakhani, K. R. (2023). Navigating the jagged technological frontier: field experimental evidence of the effects of AI on knowledge worker productivity and quality. Harvard Business School Working Paper No. 24-013.