Training / Pre-Training
Definition
The process of teaching an AI model by showing it enormous amounts of data and letting it adjust until it gets good at predicting or producing what's wanted. Pre-training is the first and biggest stage, where a model absorbs broad patterns from a vast collection of text or images. It's slow, expensive, and done long before you ever touch the tool. By the time a model reaches you, the hard learning is finished.
Why It Matters
It explains why models have a knowledge cutoff and why they cost so much to build. Training happens once, in advance. What you use afterwards is the finished result.
What does that look like in practice?
Before you ever open a chatbot, it spent months in training, working through a vast pile of text.
What people actually mean when they say this
When someone says a model was 'trained on the internet,' they mean it absorbed patterns from huge amounts of text during pre-training, once, in advance.
Last reviewed: June 2026
Related reading
From the Unspun Blog
- U.S. data centers could account for about 20% of national electricity consumption by 2035 — and that already touches bills on some gridsBNEF projects U.S. data centers could account for about 20% of national electricity consumption by 2035. What that means for bills and regional grids.
- Why Does AI Make Things Up, and Can I Actually Trust It?AI tools make things up. It's called hallucination, and it's not a glitch. Here's what's actually happening and which tasks are genuinely safe.
Knowing the word is just the start.
We help non-tech people go from looking things up to actually using AI confidently. Tutorials, prompts, plain English. All of it for $7 a month, and the first seven days are free.
Show Me How This Works →Start free. Cancel anytime. No judgment, ever.