Is prompting a lasting skill or a passing phase?
October 6, 2026 · 15 min read
Amanda Askell and the Cursor team treat prompt writing as careful craft, Fei-Fei Li and Manolis Kellis see it as a new literacy, and Aravind Srinivas says good products should remove the need for it. The answer tells learners how much effort to invest in the skill.
The rules, numbers and steps worth keeping in front of you.
- Wording tricks are likely to fade; clear asking and good follow-up questions will last.
- Askell: for everyday tasks, just ask.
- Askell: heavy prompting is for the top 2 percent of performance or prompts at a product's core.
- Askell: read your prompt as if you had none of the context in your head.
- Askell: when the model gets it wrong, ask why and what instruction would have prevented it.
- Askell: define terms first; say exactly what counts as polite or rude.
- Askell: emphatic words in a system prompt nudge how often a behavior appears.
- Cursor: users should write naturally; the product should fetch the right context.
- Cursor: showing an example can carry intent better than explaining.
- Li: prompting is a skill worth teaching in school, in the sense of Socratic questioning.
- Kellis: asking for step-by-step thinking works on models and on children.
- Srinivas: prompt engineering won't last; a better product lets you be lazier.
- Srinivas: people are not naturally good at asking questions.
- Prompting does not fix a weak model or thin, stale source material.
This is the short form. The full insight has every expert's reasoning, their own words and the sources.