Think Together · Start with a Short Story
The robot first adjusts its method with practice cards, then answers using a new card. The girl calls the first part “training” and the second “using the model.”
Lay Out the Process
- TrainingUpdate parameters with examples
- SaveKeep the trained model
- InferenceUse the model on new input
Training changes a model’s parameters; inference usually uses the existing parameters to calculate an output.
It is a little like preparing a set of tools for solving problems and then using them on a new one. But a real model does not necessarily remember rules one by one the way a person might.
Some systems separately update memory or collect feedback. Those are still different from updating the model’s parameters on the spot.
Now You Try
If you tell AI, “My imaginary character is named Blue,” and it uses that name later, does that prove its parameters changed?
Ready? Open this to see the explanation
No. The name may simply be information from the current conversation context.
One Idea to Take Away
Training changes a model’s parameters; inference usually uses the existing parameters to calculate an output.
Try covering the diagram above and explaining each step to someone in your family. If one step is hard to explain, return to that card and look again.