AI Terminology Glossary
AI Terminology Glossary
A working glossary of key artificial intelligence terms, covering how AI models are built and trained, generative AI and large language models, the history of the field, deployment risks, and the policy and ethics vocabulary needed to follow AI regulation and public debate.
Compiled by ALT-TEXT for wiki.alt-text.eu, as a companion to the Media & Information Literacy Glossary.
Foundations: how AI models are built and trained
- Artificial intelligence (AI)
- Machine learning
- Deep learning
- Neural network
- Algorithm
- Foundation model
- Transformer (AI architecture)
- Tokenisation
- Parameters (AI model)
- Inference
- Data ingestion
- Training data
- Gradient descent
- Backpropagation
- Loss function
- Epoch
- Overfitting
- Knowledge distillation
- Quantisation
- Embeddings
- Scaling laws
- Synthetic data
- Ontology (AI and knowledge representation)
Generative AI and large language models
- Large language model (LLM)
- Generative AI
- Prompt engineering
- Context window
- Hallucination (AI)
- Fine-tuning
- Reinforcement learning from human feedback (RLHF)
- Multimodal AI
- Agentic AI
- Retrieval-augmented generation (RAG)
- Small language model (SLM)
- Zero-shot and few-shot learning
- Emergent capabilities
History and schools of thought
Deployment, safety and risk
- Prompt injection
- Jailbreaking (AI)
- Guardrails
- Model collapse
- Benchmark (AI evaluation)
- Model card
- Latency (AI)
- Observability (AI)
Trust, authenticity and media literacy
Ethics and society
- Algorithmic bias
- AI alignment
- Explainable AI (XAI)
- AI washing
- Data annotation
- Surveillance capitalism