Introduction to prompting with Mistral AI

Generative AI moved from laboratory curiosity to everyday tool faster than any technology before it, and public administrations can no longer treat it as a distant subject. This one-day training starts from zero and assumes no prior knowledge. It alternates short, plain-language explanations with hands-on workshops.

Participants first learn what happens inside a model — how machine learning depends on data labelled by people, why a language model predicts the most probable next word rather than "understanding", and how context steers its answers.

They then move to practice: simple versus advanced prompts, a five-step formula for structuring an instruction, meta-prompting, and a "week working with AI" thread covering brainstorming, analysing a report, adapting writing style, extended research, visualisations and image generation.

A dedicated section covers bias, hallucinations, data protection, copyright, transparency, the EU AI Act and practical risk-control habits. The course is aimed for all beginners with generative AI who want to use it confidently in their daily work with a special focus for public-sector staff — no technical background required, with all examples rooted in real administrative situations and practised on a tool such as Mistral. Its twofold aim is to deepen understanding of the technology and to build practical AI skills in a structured way.

Understanding Generative AI

- The Human Work Behind AI (annotation, correction)

- Predictive AI in Everyday Applications (spam filters, recommendations, Face ID)

- Large Language Models: Next-Word Prediction

AI for Administrative Tasks

- Productivity Gains — and Where AI Reduces Productivity

- A Typical AI Journey: Awareness and Training, Chatbots, Internal Data, Workflow Automation (Agents), Strategy and Use Cases

- Future of Jobs (Wharton Study)

- AI as a Colleague

Prompting Workshop #1

- Prompt Engineering Fundamentals

- Simple vs. Advanced Prompts

- The 5-Step Formula for an Advanced Prompt

- Prompting Best Practices and Practical Tips

Limits and Challenges

- Bias and Discrimination

- Hallucinations

- Data Protection

- Copyright and Intellectual Property

- Transparency and Reliability of Content

- AI Best Practices

At the end of the training, learners will be able to:

- Explain how a large language model produces its answers.

- Write structured prompts using the five-step formula and established prompting best practices.

- Apply generative AI to everyday professional tasks such as brainstorming, analysing a report, drafting in a given style and researching a topic.

- Assess the reliability of AI-generated output and identify hallucinations, bias and other errors.

- Apply data-protection, copyright and EU AI Act requirements when using AI tools at work, and identify relevant AI use cases in their own administration while judging where AI adds value and where human expertise must remain in control.

This course combines lecture-based instruction (short, plain-language explanations of AI fundamentals) with practical exercises (hands-on prompting workshops) and real-world scenarios rooted in administrative tasks. Participants engage in structured problem-solving activities covering brainstorming, report analysis, text adaptation, and research.