In this article
Long channel posts and article pages can be condensed. There is a control that produces a short version with the main points pulled out, sitting above the text you were about to decide whether to read.
Useful in one specific way and quietly risky in another, and the difference is entirely about what you intend to do with what it tells you.
What you will be able to do
- check_circle Summarise a long post or article.
- check_circle Use it as triage rather than as the answer.
- check_circle Name the parts that reliably go missing.
- check_circle Decide when not to use it at all.
The summary appears above the text rather than replacing it — the full version is still there.
short_text Shortening is choosing, and something else did the choosing
A summary is not a smaller version of a piece of writing. It is a set of decisions about which parts mattered, made by something that does not know why you opened it. Two people reading the same article want different things out of it, and a summary can only serve one imaginary reader.
Most of the time that is fine, because most of the time you are deciding whether this is worth fifteen minutes. For that question a summary is excellent and the cost of it being slightly wrong is nothing.
It stops being fine the moment the summary becomes the thing you act on. What disappears first in any condensation is the qualifying material — the conditions, the exceptions, who said it, how confident they were — and those are precisely the parts that decide whether a statement applies to you.
Deciding whether to read it, or reading it instead
Same button, two completely different activities.
filter_list As triage
fact_check As the answer
Treat it as a table of contents. It is never the document.
psychology_alt The habit that makes it safe
One small practice covers nearly everything: when a summary gives you a fact you intend to use — a date, an amount, a requirement, a name — go and find that fact in the original text before using it. It takes a few seconds because you now know what you are looking for, and it catches the specific failures this tool has.
The second practice is noticing when you are reading a summary of something you would not have read at all. That is the case where the tool is doing something other than saving time: it is producing the feeling of being informed about a piece you never engaged with, which is a worse position than not having opened it.
feed If you are the one being summarised
For anybody writing long posts, this changes how the first part of a piece works. A reader deciding whether to continue may now be reading a machine's account of your argument rather than your opening paragraph, which means burying the point stops being a stylistic choice and starts being a risk.
Two things follow. Put the conclusion somewhere early and in plain language, so that any condensation of it is still your conclusion. And keep the conditions attached to the claims they qualify, in the same sentence where you can — a caveat sitting three paragraphs below the statement it limits is the one most likely to be dropped.
settings Using it
- Use it on long posts and articles The control sits with the post.
- Ask it whether to read, not what to think That is the safe question.
- Check any fact you intend to use Against the original text.
- Never act on a summary Where money, health, law or a deadline is involved.
- Do not forward what you only skimmed You become the source.
- Writers: put the point early It is what gets condensed.
This page shortens text by machine. The curriculum step covers the other machine between you and what somebody wrote: what translation is good at, where it is fluently wrong, and when to go back to the original.
At a glance
- 1AI summaries
Summarise long posts and articles.
- 2For triage
Use it to decide whether to read, not instead of reading.
- 3Details go missing
Conditions, exceptions and numbers often drop out.
Updated
