AI bootcamp for EU affairs, day 2: your AI dictionary
A practical dictionary for EU affairs teams: understand narrow AI, general AI, machine learning, generative AI, LLMs, training data, bias and why prediction is the word that matters.
Sebastián Rodríguez Pérez is a European campaign strategist with direct experience running and advising campaigns for some of Europe's largest pro-European organisations. He founded the European Campaign Playbook to codify practitioner knowledge across the EU's 27 political environments and make it accessible to campaign professionals, advocacy teams, and civic organisations working at the European level. His work spans electoral strategy, EU public affairs, digital campaigning, and the integration of AI tools into political communication workflows.
Disclosure
Sebastián Rodríguez Pérez is the founder of the European Campaign Playbook and Campaign Intelligence Library. He has worked with pro-European organisations across multiple EU member states. He maintains editorial non-partisanship in all published content; where articles relate to campaigns or organisations he has directly worked with, this is disclosed within the article. He holds no current positions in active electoral campaigns.
What it covers
The vocabulary EU affairs teams need before discussing AI tools, regulation or organisational use: narrow AI, general AI, machine learning, deep learning, generative AI, LLMs, training data, bias, prediction and prompting.
Who it is for
EU affairs, public affairs, policy, advocacy and communications professionals who use AI tools but want clearer language for judging what those tools can do, what they cannot do, and what risks they introduce.
When to use it
Use this before choosing an AI tool, writing internal AI guidance, discussing the AI Act, reviewing AI-generated work, or joining any meeting where people use “AI” to mean several different things at once.
Key takeaway
Most AI tools you use today are narrow AI. They do not know what is true; they generate likely outputs from patterns in data. Once you understand that, you can use them more intelligently, review them more carefully and discuss regulation with more precision.
In brief
- “AI” is an umbrella term covering several layers of systems and methods.
- Narrow AI is what teams use today: tools designed for specific tasks such as drafting, summarising, classifying, generating images or assisting with research.
- General AI, or AGI, would be able to learn and reason across domains. It does not exist yet, although timelines are debated.
- A useful mental model is: artificial intelligence contains machine learning; machine learning contains deep learning; deep learning enables much of modern generative AI; LLMs are one kind of generative AI.
- Modern AI systems are powerful prediction systems. They produce likely outputs, and you supply the verification.
- Training data matters because model outputs reflect the patterns, omissions and biases in the data used to build them.
- Prompting matters because instructions, context and constraints shape what the model predicts next.
Framework
The day-two framework: the AI stack
Five layers, each inside the one before it. Many of the tools you use at work sit in the inner two.
| Layer | Label | Description |
|---|---|---|
| Layer 1 | Artificial intelligence | The broadest category: technologies that perform tasks usually associated with human intelligence. |
| Layer 2 | Machine learning | Systems that find patterns in data and apply them to new cases. |
| Layer 3 | Deep learning | Machine learning that uses many-layered neural networks to recognise complex patterns. |
| Layer 4 | Generative AI | Systems that create new text, images, audio, video or code from learned patterns. |
| Layer 5 | Large language models | Generative AI that predicts and produces language. It generates likely outputs, which you then verify. |
Day two starts with a shared vocabulary
By the end of this article
You will be able to:
- explain the difference between narrow AI and general AI;
- describe how AI, machine learning, deep learning, generative AI and LLMs relate to each other;
- understand why modern AI systems are best treated as prediction systems;
- explain why training data shapes outputs, bias and blind spots;
- use a short AI dictionary in conversations about tools, regulation and risk;
- ask better questions before using AI in EU affairs work.
The problem: people say “AI” when they mean different things
In EU affairs, the word “AI” now appears in tool demos, regulation, risk briefings, policy papers, campaign plans and internal strategy meetings. The problem is that people often use the same word to mean very different things.
One person may mean ChatGPT drafting a briefing. Another may mean an image model creating campaign visuals. Another may mean automated decision-making in public services. Another may mean future general intelligence. Those are four different conversations.
This article gives you the basic dictionary. Not so you can sound technical, but so you can ask better questions: what kind of AI are we discussing, what does it predict, what data shaped it, who reviews the output and what risks follow from using it?
Who will benefit most
This is especially useful for:
- policy officers asked to brief colleagues on AI;
- public affairs teams assessing AI tools;
- communications teams using generative AI for drafts;
- managers writing internal AI guidance;
- campaign or advocacy teams that need to separate useful AI practice from hype.
Start with the dictionary
Everything below is the vocabulary you need for the rest of the bootcamp. By the end, you should be able to explain what kind of AI you are talking about, what it is predicting and what a human still needs to check.
Every prompt from this bootcamp is collected in the AI prompt library.