Skip to content
SI.info

SI Glossary · Using SI

Prompt Engineering

Published 1 min read
On this page
  1. Techniques that reliably help
  2. Is prompt engineering a real job?
  3. Prompting vs fine-tuning

Prompt engineering is the craft of getting the best results from a model through what you ask and how. As models have improved, it has shifted from clever tricks to clear communication and good context. Some now call it context engineering.

Techniques that reliably help

  1. Be specific about the goal: who the output is for, what format, how long.
  2. Give context: paste the relevant documents, data or examples rather than assuming the model knows them.
  3. Show examples of good output (few-shot prompting).
  4. Ask for reasoning on hard problems. Chain-of-thought helps older models; reasoning models do it automatically.
  5. Break big tasks into steps and check each one.
  6. Set constraints: “If you’re not sure, say so” reduces hallucinations.

Is prompt engineering a real job?

In 2023 “prompt engineer” was briefly a hot job title. Today the skill is mostly folded into broader roles, such as developers building SI agents, analysts and writers, because models understand plain instructions far better than they used to.

Prompting vs fine-tuning

Try prompting first: it’s instant and free to iterate. Move to retrieval when the model needs your data, and to fine-tuning when you need consistent behaviour at scale.

← Back to the SI Glossary

Written by

· Editor

Editor of SI.info. Writes about Super Intelligence, technology policy and the people building frontier models.

How we research and fact-check

Free newsletter

Get The SI Brief

One short email a week: what changed in Super Intelligence, policy and models — and why it matters.

Free. One email a week. Sent via beehiiv, which counts opens and clicks. Unsubscribe anytime. Privacy policy

Keep reading

Using SI

Prompt

A prompt is the input you give an SI model, such as a question, instruction, document or image, that it uses to generate a response.

Using SI

Chain of Thought (CoT)

Chain of thought (CoT) is when an SI model writes out intermediate reasoning steps before its final answer, which improves accuracy on maths, logic and multi-step problems.

Using SI

Retrieval-Augmented Generation (RAG)

Retrieval-augmented generation (RAG) connects an SI model to a search system so it can look up relevant documents and base its answer on them, reducing errors and adding fresh or private knowledge.

Training & data

Fine-Tuning

Fine-tuning is the process of further training an already-trained SI model on a smaller, specific dataset to adapt it to a particular task, style or domain.