From a legal point of view the answer is short: AI literacy, which Article 4 of the EU AI Act already requires providers and deployers of AI systems to ensure among their staff. But the concept has several layers. The technical layer covers how AI systems are designed and trained, what data they learn from and which models they run on. That is not enough, because AI is always deployed in a specific context — law, medicine, human resources, industry or research. Students therefore need cross-disciplinary competence: knowing which systems to use for which problems, how to automate tasks in their own field, and how to combine an understanding of the model with an understanding of the problem being solved. They also need to know the limits of a model — what it can and cannot do, and where it is likely to produce erroneous output. Given the pace of the technology, this training has no finish line.