i learnt prompt engineering after realizing AI wasn’t the problem.

your job is to shape the pattern so the model has no room to be lazy, vague, or capping shit.

think of prompting as programming in English.

do this when next you use AI for better results and thank me later -

1. Clear Instructions -

a good prompt should contain details, think of it as a puzzle the more pieces you give it the better the outcome.

eg - explain blockchain. (without details)

- explain blockchain in simple terms for a non-technical student, using one analogy, in under 150 words. (with details)

2. Adopt a Persona -

models don’t have one personality. They simulate one based on context given.

give your AI a role depending on the context for the prompt.

eg - act like sam altman, act as a senior product manager.

3. specify the output format -

always try to define a format for your answers

eg - one-sentence summary
bullet list of key points

especially, wen copy pasting.

4. prevent hallucinations

Ai always wants to be helpful, If it doesn’t know, it will guess.

you can remove its permission to guess.

eg- If you are unsure, say you don’t know.
- do not assume facts not provided.

5. avoid leading the answer -

If your prompt contains an opinion, the answer will mirror it.

don’t make your prompt so leading which might indirectly tell the AI the answer you are expecting.

6. limit the scope -

if you are asking for a board topic, limit the scope to a niche.

trying to cover everything guarantees shallow output.

bonus -

7. provide examples

one example beats ten instructions.

that’s all for now.

start treating AI like something to instruct, the results will be surprisingly different.

follow for more content like this, you can’t memorize this at once so bookmark and tag a friend.

you are welcome.