Fine-tuning
Taking a general-purpose AI that already knows a lot and giving it extra training on a narrower set of examples so it becomes better at one particular job. Think of it as an apprenticeship: a capable all-rounder learns the ropes of a specific trade. This further training is called fine-tuning.
A general AI model is a bit like a bright graduate: broadly knowledgeable, quick to learn, but not yet a specialist in anything. Fine-tuning is how you turn that all-rounder into an expert in a chosen area. You give the existing model a fresh batch of focused examples and let it learn from them, sharpening it for a particular purpose.
The apprenticeship comparison fits well. A skilled worker joins a specialist workshop and, through weeks of guided practice, picks up the habits and knowledge of that trade. They did not start from nothing. They brought years of general ability and added a specialism on top. Fine-tuning adds that specialism to an AI that already has broad capability.
For you, fine-tuning explains why some AI tools feel tailor-made for a field, such as medicine, law, or customer support. Under the surface they may be a familiar large language model that has been fine-tuned on training data from that field, so it speaks the right language and follows the right conventions.
Fine-tuning is one part of the wider machine learning toolkit, and it is not the only way to make an AI more knowledgeable about a topic. Feeding an AI relevant documents at the moment you ask, through retrieval-augmented generation, is another route, and often a simpler one.