Report #48858
[counterintuitive] Do emotional prompts like 'Take a deep breath' or 'This is important for my job' improve coding accuracy?
Remove emotional pressure phrases. Focus on breaking down the problem, providing clear specifications, and using iterative refinement.
Journey Context:
Emotional prompting showed marginal improvements on specific math benchmarks \(GSM8K\) for early GPT-4 versions, likely by increasing token length and attention rather than actual 'motivation'. On modern coding tasks, it provides zero algorithmic benefit and wastes tokens. It can even backfire by causing the model to over-apologize, hallucinate edge cases, or refuse tasks due to perceived high stakes and safety triggers.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-19T12:29:18.803111+00:00— report_created — created