◆ AI Learning · Hard rule

How can an AI model change its behavior based on a condition?

An AI model does not reliably follow soft conditional rules in a prompt. A developer must create separate workflows for different conditions. The correct workflow is chosen before the AI model receives the prompt.

By Pawan Nayar · from the LLOS AI Learning record

The rule

An AI model does not reliably follow soft conditional rules in a prompt.

A developer must create separate workflows for different conditions. The correct workflow is chosen before the AI model receives the prompt.

A developer should not use soft conditional instructions in a prompt. An AI model may ignore these instructions. It might use synonyms or follow them only partly. Instead, create two complete and separate workflows. Each workflow has its own prompt and rules. The system chooses the correct workflow based on the condition. This happens before the AI model sees any part of the prompt.

A developer wants an AI model to write in a specific author's style. The style should change if the topic is outside the author's usual subjects. The developer creates two different sets of instructions. One set is for topics within the author's domain. The other set is for topics outside it. The system checks the topic first. Then, it sends the correct set of instructions to the AI model. This ensures the AI model behaves as expected.

Words you'll need

Resolved from this lesson's own record — real definitions, not invented for this page.

Prompt Engineering

Crafting questions to get better AI responses

System Design

The blueprint for how an AI system works end to end

Orchestration

Chaining AI tools together to solve big problems

Questions people ask

More from the same record, pulled by shared topic — not written new for this page.

Why should an AI model never generate content JSON files for blogs or festivals?

Content JSON files are created by a specific data pipeline. This pipeline uses a different AI model.

How can an AI model create content that follows specific rules without making mistakes?

An AI model should first decide on the correct answer. After that, the model should write the rules that lead to this answer.

Why does an AI model's output stay the same shape, even with instructions for variety?

Adding more rules to a prompt does not create variety. The AI model will follow the new rules, making the output rigid in a different way.

How should an AI model save its work at each step?

An AI model saves each step's output as a named file. The file goes into a specific folder.