The company has adopted an AI policy and launched a chatbot. But does that mean the issue of AI governance is now resolved? Not quite. In addition to the question of which types of artificial intelligence employees are permitted to use and how, another question arises: Do employees know how to use these tools competently and responsibly?
This is precisely the area in which Article 4 of the European AI Act comes into play.
The AI Act Addresses Technology and Competence
Article 4 deals with “AI literacy,” or the competence of individuals working with AI systems. Companies must take measures to develop such competence among their employees and other individuals working with these systems. When doing so, companies should take into account factors such as technical knowledge, experience, training, and the specific context in which the AI is used.
This is an important point. The AI Act does not require that every employee attend the same standardized AI training. Nor does it mandate a specific certification or uniform level of competence. Rather, the European Commission emphasizes a context- and risk-oriented approach.
A developer needs different knowledge than a marketing employee. Managers deal with different issues than people who work with an AI assistant daily. What about a requirements engineer? It’s worth taking a closer look at their situation as well.
What Does AI Competence Mean in Requirements Engineering?
Requirements engineers work with critical project and product information. They analyze documents, communicate with stakeholders, organize information, define requirements, evaluate their quality, and oversee large volumes of requirements. AI can now support many of these activities.
For instance, it can derive initial requirements from workshop notes, summarize extensive documents, rephrase requirements, and flag potential quality issues. AI can also suggest questions for stakeholder interviews and assist in classifying large sets of requirements.
However, this raises new questions: Can I rely on an answer that sounds plausible? How do I recognize hallucinations, distortions, or missing context? How does my prompt affect the result? Most importantly: Who is ultimately responsible for the requirements?
The fact that a requirements engineer can use ChatGPT or Copilot does not answer these questions.
A General AI Orientation Falls Short
This is precisely why we believe the AI Act’s emphasis on context is so important. First, however, employees need a shared basic understanding. What is artificial intelligence? How do modern language models generally work? What opportunities and risks do they entail?
The European Commission cites these fundamentals as possible starting points for AI literacy initiatives. However, the Commission then emphasizes that the role, existing knowledge, specific AI systems, risks, and context of use should be considered.
A requirements engineer does not need to be able to develop large language models. However, they should understand why a language model can “hallucinate.” They should be able to critically evaluate results. They should also know how much context an AI system requires and where risks related to data protection, confidentiality, or bias arise. Most importantly, they should understand how to meaningfully integrate this technology into their work. The AI4RE seminar addresses these needs.
The International Requirements Engineering Board developed the AI4RE Micro-Credential, a compact continuing education module for using artificial intelligence in requirements engineering. The content addresses what role-specific AI competence can mean for requirements engineers.
In this seminar, participants will explore the fundamentals of modern AI technologies, as well as the capabilities and limitations of large language models. They will learn how to formulate effective prompts and critically evaluate results. The seminar also covers risks and responsibilities, such as bias, data protection, and confidentiality. Participants then apply this knowledge to specific tasks in requirements engineering, such as identifying, documenting, validating, and managing requirements.
From Theory to Practical Application
My name is Nils Janßen, and I am a requirements engineer and trainer. I help companies implement requirements engineering in a practical and methodologically sound way. This combination of solid fundamentals and concrete application is also important to me when it comes to AI.
That’s why the AI4RE seminar at microTOOL goes beyond the official course material. Together, we’ll explore what AI support can look like in day-to-day requirements engineering work. We’ll discuss more than just general-purpose chatbots; we’ll also explore AI features in existing requirements engineering tools. This is a crucial difference. After all, a language model that knows only the content of a single prompt operates under completely different conditions than an AI that can access existing requirements, project structures, and other relevant artifacts.
Furthermore, we’ll explore designing specialized AI assistants and custom agents for specific requirements engineering tasks.
In this way, the seminar goes beyond merely using individual AI tools. The universal chat window becomes a tool that can be embedded into your requirements engineering process.
Conclusion: An AI Policy Alone Does Not Create AI Competence
The EU AI Act clarifies that when using artificial intelligence professionally, the focus should not be solely on the technology. The people working with it also play a crucial role. For companies, this means that providing AI tools and regulating their use through policy is not enough. Equally important is determining the competencies that different employee groups need for their specific applications.
The IREB AI4RE Micro-Credential offers requirements engineers a technically sound approach to building these competencies.
Are your requirements engineers already working with AI? They should know not only that they are allowed to use AI, but also how to use it in a targeted, critical, and responsible manner.