Since February 2025, an article in the European Union's AI Act has asked companies that build or use AI systems to take measures to ensure their staff have a sufficient level of AI literacy. It is one short paragraph in a long regulation, and its exact reach is still being argued over. But it points at something bigger than any one law. Knowing how the AI tools around you actually work is becoming part of the job, not a hobby.
For most engineers this will first arrive as a training module. An hour of slides, a short quiz, a certificate in the HR system. It is tempting to click through it during a meeting. Before you do, notice what it is quietly saying. Your employer expects you to understand the limits of the tools you use, and may one day be asked to show that you did.
That changes what it means to be good with AI at work. It is not only about getting a tool to produce code quickly. It is about knowing when its output needs checking, what it should never be given, and how to explain to a colleague why an answer looked right and was wrong.
So build that understanding properly, whatever the module covers. Learn roughly how the tools you use are trained and why they can be confidently wrong. Find out your company's rules on what data may be pasted into which tool, and why those rules exist. Keep a short list of real mistakes you have caught in generated code or text, because those examples teach colleagues more than any slide.
Expect this to spread beyond engineering. Product managers, support teams and lawyers will all be learning the same basics. The engineer who can explain these tools plainly to a non technical colleague, without hype and without fear, will be asked into a lot of useful rooms.
And if your company offers nothing at all, ask for something. You will be helping the business with an obligation it may not have noticed, and you will be the person who raised it.
Monday: write down the AI tools you used last week, and next to each, one thing it is known to get wrong.
– Asael Shinder
Top comments (0)