← Trooper · AI glossary

AI glossary – the concepts explained

AI concepts explained so that everyone in the organisation can join the conversation – calmly, concretely and without technical jargon. Each entry gives you the definition, an everyday example and a sentence you can use with management.

28 entriesNo technical jargonUpdated continuously

The core concepts

07 entries

Artificial intelligencesoftware that solves tasks which would otherwise require human thinking.→ Machine learningwhy software can do things nobody programmed it to do.→ Generative AIthe difference between creating and recognising.→ LLM – large language modelwhy the model writes convincingly – even when it is wrong.→ Chatbotfrom fixed answering machines to AI assistants you think out loud with.→ Multimodal AIwhen the model can read, see, listen and speak.→ AGIthe concept behind the headlines about artificial general intelligence.→ Training datathe data an AI model learns its patterns from.→ Fine-tuningtraining an existing AI model further on your own data.→ Embedding and vector databasethe number series that let AI search your documents by meaning.→

Working with it

04 entries

Promptyour brief to the machine – briefly explained.→ RAG – retrieval-augmented generationwhy the AI can suddenly quote your own documents.→ AI agentthe difference between a chatbot and an agent – and what it means for workflows.→ Agentic AIwhen AI does not just answer but acts – and what that requires in terms of mandates.→ Tokenthe small piece of text a language model reads and writes in.→ Context windowthe amount of text a language model can work with at once.→ System promptthe fixed instructions that steer an AI assistant in every conversation.→ Prompt librarycollections of ready-made prompts, which I don't recommend.→ AI literacythe ability to use AI knowledgeably, critically and responsibly.→

Content and creativity

04 entries

AI-generated contentwhen the first draft is free, the value shifts to the sender.→ Image generationfrom description to image – and from production to choices about identity.→ DeepfakeAI forgeries of voices and faces – and what they do to trust.→ Tone of voice and AIwhy your voice is a strategic choice when the machine can imitate any style.→ Synthetic mediaimages, audio, video and text created or altered with AI.→ AI labellingtelling the audience that content was made with AI.→ Editorial AI policythe rules for AI in the content the organisation publishes.→

Responsibility and ground rules

09 entries

Hallucinationwhen the model makes things up – why it happens, and what the safeguard is.→ Biasthe machine's skews are inherited from data – and they do not disappear on their own.→ Black boxwhen the system cannot explain its answer – and what that means for accountability.→ Human-in-the-loopthe human in the workflow – and why it requires time and a mandate.→ The EU AI Actthe EU's AI law in brief – and what it requires of those of you who use AI.→ Copyright and AIwho owns what the machine creates – and what may it be trained on?→ Data protection and AIwhat can you share with a language model – and on what terms?→ AI policythe organisation's shared ground rules – and why they must not end up in a drawer.→ Shadow AIthe AI use management cannot see – and why bans make it worse.→ High-risk AIAI systems used in decisions about people's lives and opportunities.→ Article 4 and the AI literacy requirementthe requirement for AI literacy among everyone who uses AI at work.→ Transparency obligationsthe rules on telling people when they meet AI.→

My models and frameworks

04 entries

The FOCUS modelmy Danish model for a good prompt – five elements.→ The context ladderfour steps from rule-based to direction-setting work.→ The AI role atlasdeliberate choices about the role of AI – task by task.→ The RISE modelRelation, Intention, Steering, Enablement – the foundation for AI adoption.→

The list is expanded continuously. If there is a concept that trips you up in your day-to-day work, write to me – and it will be included in the next round.

From concepts to competencies.

Get in touch with Loa →