AI fluency

How to Tell Where You Stand With AI at Work

4 min read Scored by AI fluency

Sample scores, not an average of real users.

Someone in your Monday meeting says they've been using AI for months. You've tried it a few times. You nod, and afterward you can't tell whether you're ahead of them or behind.

Career Lab calls this AI fluency and looks at it in three areas: choosing the work you give to AI, checking the result, and doing it the same way twice.

How does the assessment check where you stand?

The AI fluency assessment asks seven questions. Six are scored, two in each of the three areas. The seventh asks which task you do every week that an AI tool hasn't touched yet, and it isn't scored. The first question starts from your last seven days at work.

AI fluency

Start with the first question.

AI fluency

Seven questions about how you use AI at work. You get a first task to hand to AI, the instruction to paste and the check to run.

Answer seven questions

Seven questions about how you use AI at work. Answer from what you did, not what you would like to do.

How do you decide which work to give to AI?

The first question is about your last seven days at work. The answers run from giving AI nothing this week and trying it on whatever was in front of you, through using it on tasks where a mistake wouldn't matter much, to picking tasks where it saves you time and you can tell if the result is wrong. The top answer is keeping a short list of what goes to AI and what stays with you, and following it.

The second question is about what goes in. Before you put anything from work into an AI tool, do you paste whatever is fastest, stop and think only if it's obviously confidential, take out the names and numbers you know are private, check what your employer allows and then take out or replace the private details, or follow the same rule every time and use only the tools your employer has approved?

What do you do before AI-drafted work goes out under your name?

In the Forbes piece on the AI skills gap, I said this about people building AI agents: "Once participants start building these systems, they quickly realize that it still requires a human in the loop." [1] The assessment's third question asks what you do before AI-drafted work goes out under your name. The answers, lowest to highest:

  1. Send it as it is.
  2. Read it quickly and send it.
  3. Read it closely and fix what looks wrong.
  4. Check the facts and numbers against the source.
  5. Run the same checks every time, and have someone else look at anything high-stakes.

The fourth question picks one case: an AI answer includes a number, a name or a source you can't place. The lowest answer is to use it. The highest is to look it up in the original source and keep a note of where it came from. If you can't find the source, take the sentence out.

Can you do it the same way twice?

The fifth question starts with the last time AI gave you a result you were happy with. The answers run from doing nothing, because you didn't expect to get it again, to trying to remember roughly what you asked, to saving the result, to saving the instructions you used, to saving the instructions and using them again on a second task. Saving the result and saving the instructions are different answers, and the instructions score higher.

The sixth question asks what you could do if a coworker wanted to see how you use AI for one task. The top answer is giving them the instructions and the checks you run before you send the result. Write your checks under the instruction.

What do you get at the end?

The last question asks which of these you do every week that an AI tool hasn't touched yet: a first draft, a summary of something long, prep before a meeting, research or a comparison, or cleaning up or formatting data. There is a choice for none of them, if you already hand these off. Whichever you pick, your result shows a first task to hand to AI, the instruction to paste into the AI tool you already use, and the check to run before it goes out.

If at least 4 of the 6 scored questions apply to you, the result also gives you a score out of 100 and a tier, with a line for each of the three areas.

FAQ

Questions people ask

  • How do I know where I stand with AI at work?

    Look at three areas: the work you give to AI, what you do before AI-drafted work goes out under your name, and whether you can do it the same way twice. The AI fluency assessment asks seven questions across those three areas.

  • What does the AI fluency assessment ask?

    Seven questions about how you use AI at work. Six are scored, two in each of three areas: choosing the work, checking the result, and doing it the same way twice. The seventh asks which weekly task AI hasn't touched yet, and it isn't scored.

  • How does the assessment know what I do at work?

    It goes on what you report. The result says so itself: "Based on what you reported doing, not on anything we observed."

  • What if a question doesn't apply to me, or my employer doesn't allow AI tools?

    If a question doesn't apply to you, pick the last option and it stays out of your score. A score needs at least 4 of the 6 questions to apply to you; with fewer, there is no score and no tier. If you say your employer doesn't allow AI tools for your work, the result tells you there is nothing to score, and the first task it shows is yours if that changes.

  • Do I need an account?

    If you aren't signed in, you answer all seven questions first. Then it asks you to create a free account to save your score, and the form says no credit card is required.

Sources

The figures in this guide come from the primary sources below. Where a number could not be traced to its original publisher, it does not appear on this site.

AI fluency

Answer the seven questions today.

Seven questions about how you use AI at work. You get a first task to hand to AI, the instruction to paste and the check to run.