Context Setting

In AI terms, "Context" is the difference between a generic answer and a useful one. Without context, ChatGPT is forced to guess.

1. The "Pre-Knowledge" Problem

If you ask ChatGPT, "Should I hire this person?", it has no context. It doesn't know your company, the job role, or the candidate.

2. Techniques for Setting Context

Data Dumping (The "Grounding")

Paste in the relevant information using delimiters.

"Below is the job description and the candidate's resume. Based on these two documents, identify three potential areas of concern for an interview."

Historical Context

Tell the AI what has happened before.

"We have already tried increasing the budget for Facebook ads, but it didn't work. Suggest three alternative marketing strategies."

Audience Context

Who are you talking to?

• "Explain this to a CEO (Value-focused)."
• "Explain this to a Developer (Logic-focused)."
• "Explain this to a Customer (Empathy-focused)."

graph LR
    NoContext[Generic Prompt] --> GenericResult[Generic, Boring Output]
    Context[Specific Context] --> GroundedResult[Actionable, Useful Output]

3. The "State of Mind"

You can even set the context of the AI's internal reasoning.

"You are in a brainstorm session. No idea is too crazy. Be as creative and non-linear as possible."

---

Hands-on: Context Experiment

1. Ask: "What should I eat for dinner?"
2. Then ask: "I have a chicken breast, a red bell pepper, and 30 minutes. I want something healthy but spicy. What should I eat for dinner?"

Observe how the second response is actually useful, while the first was likely a list of generic guesses.

Key Takeaways

• Context = Relevance.
• Use delimiters to feed raw data to the AI.
• Always define the Audience and the Constraint.