Retrieval vs Fine-Tuning
Choose the right way to customize AI with company knowledge.
What you will learn
- Explain retrieval vs fine-tuning in clear language.
- Recognise where this concept appears in real work.
- Apply the idea and check the result using a practical activity.
The concept
Retrieval gives a model relevant information at runtime. Fine-tuning changes model behavior through training examples. Retrieval is usually better for fresh facts and private knowledge; fine-tuning is better for repeatable style or specialized behavior.
In plain English
Retrieval is like giving AI the right textbook page before it answers. Fine-tuning is like training it to answer in a certain style.
Work example
A company policy assistant should usually retrieve current policy documents rather than fine-tune on old copies.
Try it yourself
Activity
For one company use case, decide whether it needs retrieval, fine-tuning, or both. Explain why.
Knowledge check
Which use case is usually best suited for retrieval?
- A.Answering from changing internal policy documents
- B.Making every answer rhyme
- C.Changing the model's base language
- D.Removing the need for source documents
Show answer
A. Answering from changing internal policy documents
Retrieval keeps answers grounded in current documents.