Are you sure? #
People try to get better results from AI by changing models, writing longer/shorter prompts or adding more context.
A simpler technique to get better results often works surprisingly well.
When you get an answer, ask:
Are you sure?
Why it works #
LLMs frequently stop at the first reasonable answer. Asking them to verify their response forces a second pass over the problem.
I’ve seen this catch mistakes in:
- Code
- System design
- Architecture decisions
- Calculations
- Data analysis
The model doesn’t become smarter, but it spends more effort validating its answer instead of stopping at the first acceptable one.
Before spending time on a complex prompt, try adding one short question at the end of your conversation:
Are you sure?