What Is Prompt Engineering? (Fundamentals)
~10 min read
The foundational framing of prompt engineering as the 'steering wheel' for an LLM — changing instructions, not weights, to shift model behavior.
LLMs are powerful, but they don't automatically know what you want. Prompt engineering is the simplest way to control them — think of it as the steering wheel for the LLM. Small adjustments completely shift the direction of the output.
Crucially, you're not changing weights (the learned parameters inside the model). You're changing instructions, and that changes everything. It's the fastest, lowest-effort way to get better results from any model — no training run, no GPU, no waiting.
A good prompt helps the model:
- Think step-by-step
- Follow constraints
- Stay focused
- Avoid shallow answers
Everything else in this topic — Chain-of-Thought, Self-Consistency, Tree of Thoughts, and Attentive Reasoning Queries — is really just a more structured way of applying this same steering-wheel idea to get more reliable reasoning out of the model.
💬 Deep Dive with AI
Key points
- •Prompt engineering changes instructions, not model weights
- •It is the fastest, lowest-effort lever for improving LLM output quality
- •A good prompt pushes the model to think step-by-step, follow constraints, stay focused, and avoid shallow answers
- •All reasoning techniques (CoT, Self-Consistency, ToT, ARQ) are structured applications of this same idea