Évaluation de la maîtrise de l'IA

Vérifiez qui sait vraiment travailler avec l’IA, en demandant aux utilisateurs quotidiens de ChatGPT de montrer ce qu’ils peuvent accomplir plutôt que de simplement le décrire.

Réserver une démo

La confiance de plus de 5 000 clients dans le monde entier.

Xobin

AI Fluency Assessment

Marketing Ops practical test

18:42
3 / 5

Task 3 of 5 Refine a launch email with an AI assistant

Use the assistant on the right to produce a launch email for our analytics add-on. You are scored on how you direct, correct and verify the model not on the first draft.

AI

Assistant workspace

Full transcript captured for scoring
Here's a general draft covering the main features with a closing call to action.
Too generic. Audience is RevOps leads at 200-person SaaS firms. Under 120 words, no adjectives.
Revised 114 words, opens with the 6-hour weekly reporting saving and a single CTA.
Where does the 6-hour figure come from? Cite it or drop it.
It was inferred, not sourced. Replaced with the range from our onboarding survey.
Good. Now a 40-word in-app banner variant, same claim.
Banner variant ready 38 words, one CTA, claim matches the email copy.
Tighten the subject line to under 45 characters and keep the hours-saved idea.
Subject: “Get your reporting hours back” 34 characters, claim preserved.
Ship that version. Summarize what changed between draft one and now.
Type your next instruction…Send
JR

Jason Reed

Session verified · ID checked

Captured live

Prompts sent4
Refinements3
Sources verified2
Hallucinations caught1

Session activity

Task opened00:12
First prompt sent01:40
Audience constraint added03:22
Model output corrected05:08
Unsourced claim challenged07:31
Banner variant requested09:55

What we score

  • Say who the audience is and what the constraint is.
  • Correct the model instead of restarting the prompt.
  • Check any number the model gives you.