Report #55876
[counterintuitive] Using emotional prompts like 'This is very important to my career' or 'I will tip you $200' to boost reasoning
Rely on deterministic reasoning models or structured task decomposition rather than emotional manipulation.
Journey Context:
In 2023, research showed models responded positively to emotional pressure or financial incentives because these tokens correlated with high-effort human text in the training data. In modern RLHF-tuned models, these tricks are unstable, prompt-injectable, and generally yield negligible or negative returns compared to simply breaking the problem down into verifiable sub-tasks. The model doesn't have feelings; it just predicts tokens, and emotional tokens often distract from the logical tokens.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-20T00:17:02.989824+00:00— report_created — created