"Proficient in AI tools" is now on so many CVs that it has stopped carrying information. It costs one line to write and nothing to back up, so the people reading applications have quietly stopped treating it as a signal.
That is not cynicism. It is what happens to any claim that is free to make.
What the check actually looks like
When a claim matters, someone tries to confirm it. In practice that is one of four things:
- A question in the interview. "Walk me through the last thing you used AI for at work." Vague answers end the topic.
- A small task. Draft this reply, summarise this document, clean up this list. They are watching your process, not your typing speed.
- A reference. Did this person actually do the work described?
- A credential. Something with an issuer and a code they can check themselves.
Only the last one survives a check the candidate is not present for. That matters, because most rejections happen before anyone speaks to you.
Why the task trips people up
The small task is where "I use ChatGPT every day" separates from "I work with AI."
Daily chatting builds familiarity with a text box. Work requires something else: knowing which tasks to hand over, briefing with enough context to get a usable first draft, and (the part that gets missed) checking the output before it reaches someone who is relying on it.
That last habit is the one hiring managers are most quietly testing for. A confident wrong answer that reaches a client is worse than no answer at all, and everyone who has worked with these tools for a while has a story about it.
Making the claim checkable
The fix is not a longer CV line. It is making the claim survive a check you are not in the room for:
- Be specific about the work, not the tool. "Rebuilt a client's weekly reporting so it takes 40 minutes instead of three hours" beats any tool name.
- Keep one example you can walk through. The prompt, what came back, what you changed, why.
- Carry proof that stands on its own. A credential with a public verification page answers the question before it is asked.
None of this requires you to be an expert. It requires the claim to be the kind of thing someone can confirm, which is a much lower bar than being the best, and a much higher bar than writing a line.
Where to start
If you are not sure which of these you can already do, the AI Readiness Test scores you across prompting, tool choice, workflow, and judgment in about ten minutes, and tells you which one is weakest. It is free and there is nothing to install.