Training
Training is the process of showing a model vast amounts of data and repeatedly adjusting its internal numbers until its predictions improve. It's expensive, slow, and happens once — when training ends, the model is frozen and no longer learns from new conversations. What you interact with in any AI product is the result of training that already happened, sometimes months earlier.
Videos explaining this concept
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.
E006Notes on AI
Training vs Using a Model
Training and Inference are the two distinct phases of an AI model's lifecycle.
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.
E017Notes on AI
"Forgetting" vs "Never Knew"
When AI doesn't know something, there are only three structured causes: a training gap (the information was never in the dataset), an injection gap (it exists but was never added to the system), or a visibility limit (it fell outside the context window mid-conversation). In all three cases the model isn't being unintelligent — it's operating within structural constraints. Understanding which cause applies tells you exactly how to fix it.
E019Notes on AI
Why AI Sounds Confident
LLMs are optimised to produce fluent, grammatically correct language — not to verify whether what they say is true. The model generates the most likely continuation token by token, and when those tokens form smooth professional prose, the result sounds confident even if the underlying content is wrong. This is fluency bias: our brains mistake the quality of language for the quality of information. When evaluating an AI answer, ignore tone — look at evidence, sources, and consistency.