AI Critical Editing & Verification Assessment

Name: AI Critical Editing & Verification Assessment
Brand: Selection Lab
Availability: InStock

The first generation of AI mistakes came by email: blindly forwarded ChatGPT output with invented case law, hallucinations in due diligence, people promising things to clients that were untrue. The next generation of mistakes is more subtle: small numerical inaccuracies, wrong tone, an adjective in the wrong place. The AI Critical Editing & Verification Assessment measures in 20 multiple-choice questions whether a candidate can read AI output the way an editor reads a submitted manuscript: critical, fast and focused on what needs to be red-lined. For anyone who forwards AI text to clients, the public or managers, indispensable.

Time

~15 minutes

Type

Hard skills test (AI verification)

Form

Multiple choice (20 questions)

Perfect for:

Lawyers and legal professionals, Copywriters and editors, Journalists and communications staff, Consultants and analysts, HR and communications teams, Researchers and KM specialists, All roles where AI output leaves the organisation

View →

Smiling woman looking at a smartphone while wearing a gray sweater indoors.

Rounded square shape with a gradient from blue on the left to orange on the right.

3 Reasons we love this assessment

Recognising hallucinations

Recognising invented sources, made-up court cases and impossible citations is a distinct skill. The test checks whether a candidate spots an AI fabrication before it flows into a writ, article or report.

Checking numbers and context

AI easily confuses years, swaps amounts and draws unfounded conclusions from correct data. The test measures whether a candidate reads the numerical side critically instead of blindly trusting the model.

Tone, register and brand

AI writes flat and generic by default. For a law firm, a luxury brand or an activist NGO, that's not good enough. The test checks whether a candidate can feel when the tone needs adjusting.

Information

Four types of AI mistakes

Factual inaccuracies, fabricated sources, wrong logic, and tone mismatch. The test shows examples of each and measures which mistakes the candidate spots and which slip through.

Fast or thorough

Not every AI output deserves the same level of editing. The test gauges whether a candidate can judge when a quick check suffices and when you need to review top to bottom.

Source verification

AI happily produces URLs that don't exist or citations that don't hold. The test checks whether a candidate knows which tools and workflows to use for quick verification.