Series · 5 parts · 3 min
AI Fundamentals
Most explanations of AI start with the tools. These five videos start underneath them: what a model is doing when it answers you, why it sometimes gets things wrong, and what the words tokens, embeddings, attention and transformer actually mean.
Watch them in order. Each one is under a minute and builds on the last, so by the end the T in ChatGPT makes sense, and so does most of what you read about AI.
-
0:37 How LLMs think
Why AI models don't reason or search a database. They calculate probabilities to predict the next word, one token at a time.
-
0:34 How words become tokens
Tokenisation explained: the first thing that happens to your text inside an AI model, and why it matters.
-
0:39 Understanding embeddings
How AI models turn token numbers into coordinates with real meaning, and why distance between ideas powers modern search.
-
0:42 Attention explained
How self-attention lets words talk to each other and update their meaning based on context.
-
0:46 Transformers explained
What a Transformer actually is, in plain English. It reads the whole sentence at once, then every word checks every other word, over and over.