AI jargon, translated
The What-the-Hell-Did-He-Just-Say? Vocabulary List
Translate the nonsense. Make it make sense. Your jargon code key: 25 AI terms you keep hearing, explained in plain English.
Why you should care
Half the reason AI feels like a club you are not in is the vocabulary. Salespeople and the kids toss these words around to sound smart and make you feel slow. Learn these and you stop nodding along, start asking the right questions, and nobody sells you a "revolutionary AI agent" that is really a chatbot with a nice haircut.
- 1. LLM
- That's just the fanciest autocomplete you've ever seen, finishing your sentences like it already knows what you're thinking.
- 2. Token
- Those little word pieces the machine actually counts... and yes, they charge you for every single one.
- 3. Context Window
- How much it can remember before it completely forgets who you are and starts over.
- 4. Hallucination
- It just made the whole thing up and delivered it with full confidence, like it knows exactly what it's talking about.
- 5. RAG
- Look it up first, then answer. Stop guessing and go check the source before you open your mouth.
- 6. Agent / Agentic
- This AI doesn't just talk... it actually gets up and does the work for you.
- 7. Fine-tuning
- Extra training so it stops being a generic know-it-all and gets good at your specific thing.
- 8. Parameters / Weights
- All those millions or billions of tiny knobs they twisted while it was learning.
- 9. Training / Pre-training
- The expensive phase where it read the entire internet and learned every pattern.
- 10. Inference
- The actual answering part. Way cheaper once all that expensive learning is already done.
- 11. Prompt / Prompt Engineering
- What you type. The "engineering" part is just getting good at asking so it stops giving you garbage.
- 12. Embedding / Vector
- Turning words into numbers so the computer can measure exactly how similar two things are.
- 13. Transformer
- The architecture that made modern AI stop sucking. Attention is the whole game.
- 14. Neural Network
- A bunch of math layers loosely inspired by brains. Not actually a brain, though.
- 15. Generative AI
- It makes new stuff (text, images, whatever) instead of just labeling the old stuff.
- 16. Alignment
- Trying to make the AI do what humans actually want instead of what we accidentally told it.
- 17. Bias
- Garbage in, garbage out. If the training data was skewed, the output is going to be skewed.
- 18. Zero-shot / Few-shot
- It does the task with zero or just a few examples instead of needing a whole new training run.
- 19. Temperature
- The creativity dial. Low is boring and safe. High is wild and unhinged.
- 20. Multimodal
- Can handle text, images, and audio at the same time without missing a beat.
- 21. Diffusion
- How most image generators work: start with pure noise and slowly clean it into a picture.
- 22. Open Weights
- You can download the actual model and run it yourself instead of just renting it from the big companies.
- 23. Frontier Model
- The current biggest, newest, most expensive ones from the top labs.
- 24. Guardrails
- The rules and filters so it doesn't start saying insane or illegal stuff.
- 25. AI Slop
- That low-effort AI garbage flooding the internet like nobody asked for it.
Ready to use it? Read our AI tool reviews or browse the AI tool directory.