Prediction
When you send a message to an AI, it doesn't look up an answer — it predicts which word should come next, then the next, and so on. This prediction is based on patterns the model absorbed during training. Everything AI generates, from a short reply to a full essay, is the result of millions of these small predictions chained together.
Videos explaining this concept
E002Notes on AI
Why Generative AI Feels Different
Traditional AI classified and ranked things — it always felt like a system with predefined paths. Generative AI feels different because it creates: you express an intent and the system generates something that didn't exist before, rather than selecting from existing options. This shift from selection to creation is why GenAI triggers a response that feels magical or even unsettling — creation was previously something only humans did.
E003Notes on AI
How AI Thinks
LLMs are prediction engines, not reasoning systems — they predict the next token based on everything they've been trained on. The difference from human thinking is not the mechanism but the grounding: AI is grounded in the textual representation of the world, humans are grounded in lived reality. AI knows "about" everything but experiences nothing — and that distinction matters more than most people realise.
E004Notes on AI
Why GenAI Advanced All at Once
Common perception links GenAI primarily to chatbots. However, GenAI is a broad family of models sharing a single "DNA": Prediction.
E005Notes on AI
What Is a Model, Really?
A model is a file — a large set of numbers called weights that crystallize everything learned during training. The model doesn't store sentences or facts; it stores statistical patterns about how tokens follow each other. Training collapses into this artifact once, and once the file exists, copying it is cheap — which is why model security matters and why you can run an open-source model with no internet connection.
E010Notes on AI
The 5-Sentence Mental Model of GenAI
This episode provides a checkpoint after the foundational episodes, compressing the key concepts into five memorable sentences that serve as a mental compass for AI.
E016Notes on AI
Why Long Chats Drift
Long conversations degrade AI output because the context gets overcrowded — not because the model loses intelligence. As a conversation grows, instructions get diluted, topics blend together, and eventually early content falls outside the context window entirely. The fix is not to argue harder or add more text: open a new conversation, reintroduce only what matters, and clarity returns.