Here's an article about x: <text of sample article>

A quick summary of the article, focusing on the main relevant points and keeping critical detail: <sample summary>

as an example, then duplicating that with the real article and text.

That's the "few shot learning" from the original paper. It could also be that it's good enough at summarizing specifically that you don't need an example, just the right framing prompt around the article text. Either way, that kind of prompt engineering is how you get "text completion" to perform basically any text processing task, generate code or play tic tac toe, so on

You can give it anything - or nothing - and ask for any kind of output.

The relevance and accuracy of what it gives you can vary lots. You can give it examples of the types of things you want back etc. You don't necessarily have to give it examples either, just ask it a question. It normally performs better on semi-complex tasks when you give it examples though.

This is a fully general technique. Any repeatable problem can be rephrased as a prompt + related technique via the following structure:

  Problem: <Example of Problem>

  Solution: <Example of Solution>

  <...maybe include one or two more examples>

  Problem: <Real Problem>

  Solution: <Ask GPT-3 for the solution>