gemma@briefings:~/posts$ open measuring-ai-competence
Measuring AI competence: the real skill is using AI correctly and ethically
AI literacy has been mandatory since February 2025 (Art. 4 EU AI Act). But how do you measure it? Four questions anyone can start with themselves.
Using AI has long been part of everyday work. We have it draft emails, write code and summarise meetings. Building your own agent with Claude or ChatGPT takes an afternoon today. That's great, but it's no longer a differentiator. Saying "I use AI" today says about as much as "I can google".
The real competence lies elsewhere: using AI in a way that lets you stand behind the result. It's good practice, and since February 2025 it's also the law. Under Article 4 of the EU AI Act, companies must ensure that their employees have sufficient AI literacy. Which leaves the question: how do you measure that?
The easiest place to start is yourself.
1. Can you explain the result?
When a model delivers SQL, an analysis or a text, could you defend every line in front of a client? Pass AI output on unchecked, and you've handed your responsibility, along with the review work, to whoever receives it.
2. Do you know where your data ends up?
Not every piece of information belongs in every tool. At Gemma, client data only runs through infrastructure we control ourselves. If you can't say what happens to a prompt after you hit send, you have a problem, even if the result looks good.
3. Do you know which decisions build on your result?
AI models pick up biases from their training data. In an email, that doesn't matter. But if an AI-generated customer segment decides who gets a discount, or an analysis feeds into a personnel decision, "looks good and/or plausible" isn't enough. Then acting ethically means asking two questions: is the result correct, and is it fair? Who is affected, who could be disadvantaged, and who is responsible if it's wrong?
4. Do you say when AI was involved?
Transparency costs nothing and builds trust. A sentence like "Drafted by AI, reviewed for substance by me" is enough.
Counting agents or hours saved only measures part of AI competence. The bigger part is in the four questions above: do you understand the result, do you know where your data goes, have you thought through the consequences, and are you open about it?
The question is no longer "Do you use AI?" but "Do you take responsibility for the result?"
This text was outlined with AI support and reviewed and revised by Bianca Frost.
Want to know what this means for your company? 30 minutes with us, free of charge.
Book a call Learn more