26 letters. Every AI word you keep hearing.
Short, clear explanations. This is how A2Z AI explains things every morning.
A
Agent
An AI that can take steps on its own to finish a task, not only answer.
B
Benchmark
A standard test used to compare AI models.
C
Context window
How much text an AI can read and remember at once.
D
Diffusion
The method many image generators use to turn noise into a picture.
E
Embedding
Numbers that capture the meaning of text, so similar ideas can be found.
F
Fine-tuning
Extra training on your own examples to get better at one job.
G
GPT
The family of models behind ChatGPT.
H
Hallucination
When an AI states something false with confidence. Always check facts.
I
Inference
The moment a trained model produces an answer for you.
J
Jailbreak
A trick prompt that tries to make an AI ignore its safety rules.
K
Knowledge cutoff
The date after which a model has no training data.
L
LLM
Large language model: AI trained on huge amounts of text.
M
Multimodal
AI that works with images, audio or video, not only text.
N
Neural network
Connected math functions that learn patterns from data.
O
Open weights
A model anyone can download and run themselves.
P
Prompt
The instruction you give an AI. Clear prompts get better answers.
Q
Quantization
Shrinking a model so it runs faster on smaller devices.
R
RAG
The AI looks up relevant documents first, then answers using them.
S
System prompt
Hidden instructions that set how an AI assistant behaves.
T
Token
A small piece of text. AI models read and charge by tokens.
U
Upscaling
Using AI to make an image or video sharper and larger.
V
Vector database
A database that searches by meaning, not exact words.
W
Workflow automation
Linking AI with your apps so tasks run on their own.
X
Explainable AI
Methods that show why an AI made a decision.
Y
Your data
What you paste into AI may be stored. Check privacy settings first.
Z
Zero-shot
Asking an AI to do a task without giving it examples first.
