See the Difference
Same question. One dives straight in. One asks for the underlying principle first.
Adding salt raises the boiling point of water. This is called boiling point elevation. With typical cooking amounts, the increase is small — about 1°F per tablespoon of salt per quart of water.
Correct, but shallow. You get the fact without the understanding.
Principle: When you dissolve anything in a liquid, it lowers the vapor pressure. The liquid needs more energy (higher temperature) to reach the point where it boils. This is a "colligative property" — it depends on the number of dissolved particles, not what they are.
Applied: Salt (NaCl) splits into two ions in water, so it has a stronger effect than sugar. Each tablespoon raises the boiling point by about 1°F per quart...
Now you understand why, not just what.
Why This Works
When AI jumps straight to a specific answer, it often retrieves a fact without engaging deeper reasoning. By asking for the general principle first, you force it to access the underlying concepts that govern the answer.
This mirrors how experts think. A physicist doesn't memorize every specific case — they know the principles and apply them. When you ask AI for principles first, it reasons from a stronger foundation and gives more accurate, more complete answers.
How to Prompt It
- Two-step: "First, what are the general principles of [topic]? Then answer: [specific question]"
- Science/tech: "What physical or chemical principles apply here? Then tell me..."
- Decisions: "What are the key principles for [domain]? Now apply them to my situation..."
- Learning: "Explain the underlying concept first, then walk me through this specific example."
Step-Back Questions by Domain
- Physics/Chemistry: "What principles govern this phenomenon?"
- History: "What broader historical patterns or forces were at play?"
- Business: "What are the fundamental economics or market dynamics here?"
- Health: "What biological or physiological principles are involved?"
- Legal: "What legal doctrines or principles apply to this type of situation?"
The Technique
Before asking your specific question, ask AI for the general principles that govern the answer. This one extra step grounds the response in deeper understanding and improves accuracy — especially for science, reasoning, and knowledge-intensive questions.
When to Use This
- • Science and technical questions where principles matter
- • Questions that require reasoning, not just recall
- • When you want to understand why, not just what
- • Complex questions where details might lead AI astray
- • Learning new topics where foundational understanding helps
When to Skip This
- • Simple factual questions ("Who won the 2020 election?")
- • Questions already about principles ("What is Newton's third law?")
- • Creative writing or brainstorming
- • Common knowledge that doesn't need deeper grounding
- • When speed matters more than depth