Reflection: The Model Didn't Get Worse, My Prompts Got Lazier
ClickOk5811 · reddit · 2026-08-19
After blaming a model for a string of bad outputs, the author reviewed their history and realized the issue was their own increasing laziness. Early requests spelled out constraints explicitly, but later requests assumed the model would infer them from context established messages ago. The constraints that stopped appearing in the output were exactly the ones the author stopped stating.
More from Apps
- Claude Launches Gmail and Google Drive Integration with Auto-Reply — claudeai · 2026-08-19
- Feldar 1.0.1 Released: Natural Language Canvas Editing and Manuscript Import — almmaasoglu · 2026-08-19
- Grok Bot updates: Grouped notifications, voice dictation, and plugin improvements — mark_k · 2026-08-19
- Using OpenAI to aggregate health data and aid decisions after brain surgery — RishiBommasani · 2026-08-19
- TAO.com wallet launches with fiat on-ramp and Bittensor subnet trading — 0xSammy · 2026-08-19
- User prefers ChatGPT Desktop priority view, ignores projects and pins — Dimillian · 2026-08-19