It’s a rough time for those of us who have always used en- and em-dashes and who structure documents with bullet points.

I’m fortunate that writing comes easily and quickly, but these days probably most people who read my docs assume it’s AI. I can’t bring myself to abandon my beloved em-dashes though.

That people assume proper use of punctuation and clear writing are signals of AI just tells you about the sorry state of education. And it’s only going downhills from here.

I’m torn here — first instinct is to agree. But then maybe we’re just prescriptivists complaining about the kids not doing language right, like when people insist that some archaic meaning of a word is the “real” one. Maybe it’s just evolution of language and AI is writing in an older style because of training data?

You can tell I’m not AI because I use the british-style parenthetical separator ` – ` (space en-dash space) instead of the American em dash. It’s subtle, but even the smallest acts of rebellion count.

Is that a British vs American thing? I had always thought it was a newspaper vs Novel thing.

I’ve started writing “--“ and not letting autocorrect change it to be “—“

Perhaps this time of adversity will turn a bunch of formatting and type nerds into e e cummings - https://poets.org/poem/r-p-o-p-h-e-s-s-g-r

Happening the same here too. I usually write to the point and regularly number or bullet my key points. I also love to use “–”, Oxford comma, and curl all of my quotes. Now, I’ve to prove that I’m not using AI with my emails.

If AI regularly uses the Oxford Comma, it can't be all bad.

It thinks in Comic Sans, however.

The other day I wrote a long (a couple paragraphs) iMessage to 2 friends. I used bullet points ("-") in it and it was detailed. After I sent it I got the response "Did AI write this?" to which I was rather annoyed. Not only that a friend of mine would think I would send them AI Slop is a private group message but also that somewhat implies "I don't think you wrote this".

Now I know they were partly joking and they are the type of person that might not see sending AI-generated response as rude/distasteful (like I do) but it did make me sad for the state of the world. I'm sure many people just slap AI slop into all sorts of conversations with people they call "friends", I don't get it. I use AI for things I don't want to do, talking to my friends is not one of those things.

It's a font that morphes the text fragment "am-" into the curly dash using a ligature.

https://chriscoyier.net/2025/05/10/the-am-dash/

But that won't help for long. If the Amdash becomes popular, then AI will pick it up, because "am-" is just a text fragment and AI can learn to produce it.

Even worse if a future version of Unicode adopts the Amdash, then nothing will stop AI.

Or the opposite happens as one already said here: Nobody will use the Amdash.

According to Unicode's tenets, if it was used to any great degree then they are obligated to support it, are they not? Their mission is to make it possible to encode any writing glyph humans have ever used to communicate in writing or so I thought.

Yes, there's an obvious negative feedback loop to the effectiveness of this. The designer can't have realised that all symbols are just as opaque as any others to an LLM.

That site is really hard to read. I was about to say it needed a designer, but it obviously had one. But the designer is either very bad or intentionally made the design very bad. I can't tell. I also can't tell if the author is serious or being satirical.

And that flashing image at the bottom could be dangerous for people with seizure disorders.

The end of the page needs an "Epilepsy ⹃ flickering lights" warning. Looks like the Loki Disney+ series intro, in cassette futurism style.

Interesting concept that attempts to solve a problem I wasn't aware of, AI's overuse of the Em-Dash. I'm curious about other markers to proof human authorship.

spelling and grammar mistakes

You can easily prompt ai to do that

and you can easily prompt ai to use a hyphen

lowercase, spelling and style is still an indication of human writing

The styling of this reminds me of faces that cant and/or serif the hyphen. I checked a few, and in my very limited survey (Blado, Poliphilus, Trajanus and "related" faces for each that Monotype recommends) most type faces that style the hyphen do not change the appearance of the emdash at all from a horizontal line. The few counter-examples I found are italic faces (Balladeer by URW.)

If you're going to use a special font, why not use a font that completely scrambles the alphabet? That would actually be a challenge to AI

Interesting point, though perhaps a nightmare for screen readers and accessibility.

> The am dash, with its pointed unusability by AI, serves as a subtle watermark of presence—a fingerprint smudged on the edge of a sentence.

They missed using the am-dash in this sentence and instead used an —

But this punctuation can only be used by common people when it’s included in other common fonts. Then it will also suffer from AI use over time and someone will invent yad-dash (yet another dash).

You can't solve the problem of malicious actors by asking them not to act maliciously.

The theory is that it isn’t present in the corpus of training data, so AI won’t use it. Maybe sort of kind of true, but probably more performance art than actual proposal.

In the unlikely event there was uptake, websites would regex emdashes into amdashes to appear human, AI would train on that, and we’d be looking for a different secret signal.

I believe it's because the actual text behind the font would say "Here's a statementam-and here's a diversionam-and back to the original statement."

Problem is, it's only useful if you're using those specific fonts. Now I've gotta teach my grandpa how to install a font and switch to using it on every single device and application.

I think it's because it's a font ligature rather than a code pointam-it wouldn't be readable to an AI.

This is not a technical solution to AI slop. It's an artistic expression of frustration with AI slop. The fact that it isn't a solution can be read as part of the artistic objection that there doesn't seem to be a solution to the problem.

It's a joke?...

There are two possible outcomes:

1. Nobody will use the Amdash

2. AIs will pick it up and learn to use it

You might be taking this for more than it is -- at best a joke, at worst a (strikingly effective) publicity stunt.

There is the possibility that informally and in certain contexts people agree to this use this as a sign of human writing. Obviously malicious actors in malicious contexts will not follow such a practice, but it might be useful in environments of high trust.