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.

12 questions on this insight. Answer in your head, then open the card.
Is prompting a lasting skill or a passing phase? Show answer

Both. The wording tricks are likely to fade. Knowing what you want, saying it clearly and asking good follow-up questions will keep their value.

How does Amanda Askell write a serious prompt? Show answer

She defines her terms, probes the model, hunts for edge cases where it misunderstands, and revises, sometimes hundreds or thousands of times.

What does Askell tell ordinary users to do? Show answer

Just ask. Heavy prompting matters only for the last bit of performance or for prompts that are core to a product.

What does Askell mean by empathy for the model? Show answer

Read your prompt as if seeing it fresh, with none of the context in your head, and add what is missing.

What does capitalized emphasis in a system prompt actually do? Show answer

It is a nudge. It shifts how often a behavior appears, for example from 40 percent of the time to 80 percent.

Why does the Cursor team build an elaborate prompt system? Show answer

So users do not have to. Users should write naturally while the product retrieves and ranks the right context.

What does Fei-Fei Li mean when she says schools should teach prompting? Show answer

The Socratic method, seeking truth by asking questions. She says Socrates would be humanity's best prompter.

Why does Li want everyone to learn to use AI? Show answer

Agency. By learning, a person feels more in control of the technology.

Why does Aravind Srinivas say prompt engineering won't last? Show answer

A better product should understand the user and allow them to be more lazy, not less. The user is never wrong.

What problem does Srinivas still see in users? Show answer

People are not naturally good at asking questions, so his product suggests follow-up questions under each answer.

Where do Li and Srinivas really differ? Show answer

Li thinks the effort of learning to question is good for the learner. Srinivas thinks removing effort is the point of a product.

What does prompting not fix? Show answer

A model that is not smart enough for the question, or source material that is thin, stale or irrelevant. You still have to judge the answer.

This is the short form. The full insight has every expert's reasoning, their own words and the sources.