I used em dashes before it was uncool (or: writing better with AI without the slop)

I power commencement talking a small concerning myself.

Before joining Hostinger, I spent six years at a Brazilian tech news website called MacMagazine, and in the newsroom I was known as the wordy one. I was never satisfied alongside lone giving the news: I continually added my opinion, a individual angle, a bit of humor… and many of em dashes to clasp it all together in the identical sentence. Sometimes em dashes inside em dashes.

The dashes became a operating gag among my colleagues, and my editors kept asking me to go easier on them. Bear in intellect that was before AI was a item and, therefore, before em dashes became the indication for “a chatbot wrote this.”

So whenever AI penning took complete the world, I had to let go of my cherished em dash so I wouldn’t audio akin a chatbot. I mean, I never did audio akin one (I think!), but an untrained eye power have idea so.

Cut to a few years later. Now I use AI all day to assistance me compose the tutorials and blog posts for Hostinger, and alongside the way I’ve arrive to think that much of what we call AI slop started as nothing additional than individual habits, akin my dashes. AI fair has all of these habits at once, and none of the reasons rearward them. That’s what slop is to me: penning on autopilot, anywhere nobody decides which ones really belong.

Writing, following all, is thinking before it’s item else. And as much as AI can assistance me plan, research, draft, and correct, it shouldn’t substitute my thinking; whenever it does, it shows up on the page.

This extremely part you’re study was written alongside AI’s help. And I’ll display you exactly anywhere that assistance lands.

Slop is a habit without a reason

To be clear, my dashes were never the issue per se. I had a logic for all one, equal the ones my editors cut. In Portuguese, the em dash, called travessão, is mundane symbols for conversation and asides, so quarry came alongside the language.

A habit is a move you keep making since it worked before, lengthy following you’ve stopped checking whether it fits. AI picked up thousands of ours from the penning it learned from, and it reaches for them by default, without asking whether the declaration in forefront of it needs them. That’s whenever a habit turns into slop.

Openings are a fine example of this deficiency of judgment. Brazilian journalists have a name for an intro that circles its topic before landing on it: nariz de cera, or wax nose. Folha de S.Paulo, one of Brazil’s most influential newspapers, calls it a sign of wordiness that has no location in journalism. It wasn’t continually considered bad writing, though: for decades, it was merely how Brazilian newspapers opened a story, until the guide took its location in the 1950s.

I asked multiple of my reporting professors why it’s called that, and none of them could completely explain it. The finest conjecture I’ve established since is that a nose made of wax has nothing holding it up, fair akin a narrative alongside no lead.

In my experience, an AI intro written without direction is a wax nose additional frequently than not: a paragraph on why the topic matters before the topic shows up.

(If you noticed this article’s archetypal row clears its throat a little: guilty. I kept it on purpose, since it’s the one throat-clear in current alongside a reason. It tells you a individual is talking, and a wordy one at that.)

Endings have their own version. When I wrote concerning the AI to-do aide our Head of Content built, the AI outline closed on her row concerning time, afterward added one more: “And her aide motionless hasn’t sent a sole message. That’s the point.”

Ana Hercigonja Galić, one of my editors at Hostinger, who reviewed the piece, suggested cutting it, and she was right: the citation had already done the activity and had its importance to near the story. The additional row lone told readers how to awareness concerning it. A punchline to near a paragraph plant in an op-ed, but following a citation that already lands, it’s one row too many.

What AI does improved than me

The correction runs the another way, too. AI does three things improved than me: keeping things short, polishing my English, and turning a vague idea into a tangible one.

I lean to compose additional than I need to province my point, and AI is fine at suggesting shorter versions that readers can get through quickly. That matters a lot whenever they arrive to a tutorial to get item done.

Graciliano Ramos, one of Brazil’s awesome writers and a newspaperman himself, had a regulation for this. “One should compose the way the washerwomen of Alagoas do their work,” he stated in a 1948 interview. “They wet the clothes, wring them, soap them, and wring them again until not a sole autumn falls.”

“Words weren’t made to decorate, to radiance akin counterfeit gold,” he went on. “Words were made to say.” (The translation is mine.) That’s a job AI and I now share: sometimes it wrings out my drafts, and sometimes I wring out its.

The finest example of the two operating together came from that to-do aide piece. Our Head of Content’s two AI assistants can’t conversation to all another directly, so they continue notes through a shared Google Doc. My outline stated item akin “colleagues operating from extremely distinct places.” The AI tightened it to “coworkers who sit on distinct floors.”

The idea was mine, but AI made it concrete.

This part had an interviewer

I didn’t scheme this part alone. Before penning a word, I asked my AI aide to discussion me concerning how I compose alongside AI. It asked one inquiry at a period and followed my answers anywhere they went, fairly than sticking to its list, fair akin a newsman would.

It was a fine interviewer. When I told it concerning my em dashes at MacMagazine, it answered alongside three words: “That’s your hook.” It was right. It compared AI drafts of my published pieces alongside the final versions and asked which changes were mine. When I sent it the slides of a studio I gave my squad in Vilnius final year, it spotted nariz de cera in the transcript, garbled by the transcription tool, and asked me whether that’s what I’d said.

It additionally got things wrong. It archetypal thrown this part as an editor’s guide to spotting slop. “I’m not an editor, I’m a writer,” I told it. It additionally dated the term at 70 years old alongside total confidence. According to media historian Cristiane Costa, the wax nose was already the norm in Brazilian newspapers in the 1920s, so the habit is at smallest a hundred years old.

Halfway through our interview, I warned it that AI tends to be too literal and not extremely subtle, and that any idea I gave it would arrive rear a bit too in-your-face. It accepted and suggested leaving the lines that matter most to me. That’s why the archetypal and final lines of this part are entirely mine, during the remainder is a mix.

Here’s how AI suggested the beginning of this article:

You’ve already peruse what I did alongside it: identical facts, affirmative a confessed throat-clear, an “(I think!)” aside, and a screenplay-style “Cut to a few years later.” While the AI wrote a tidy opening, I wanted one that echoed akin the wordy guy from the newsroom.

Turning my habits into rules an AI can follow

After the interview, I had a catalog of everything I normally fix by hand, and I wanted to see whether an AI could keep those reasons from one conversation to the next, so I tried it alongside Hostinger Agent. It lets you preserve up to 4,000 characters of instructions as a tradition accomplishment and use them in any chat.

My skill, named “Write akin Bruno”, has eight rules, and you’re greeted to pilfer them:

• No throat-clearing. Open alongside the reader’s problem, a tangible scene, or the answer. Never open alongside a definition, “In today’s…”, or a row explaining why the topic matters.
• No disclose lines. Cut “The amazing part is…”, “Here’s the thing”, “The instruction is clear”, and “not X, but Y”.
• No kicker. Don’t add a row following a powerful citation or fact. If the matter already lands, stop.
• A logic for all step. “Do X” becomes “Do X, since Y”.
• Context for outsiders. Explain item a audience exterior the business wouldn’t know, the archetypal period it appears.
• Short paragraphs. Two to four sentences each, and no em dashes unless the declaration really needs one.
• Sharper, not safer. When tightening my writing, keep my images, humor, and different phrasing. If you’d cut one, ask me first.
• No invented facts. Never invent facts, quotes, or numbers. Mark gaps as [NEEDS SOURCE].

Then I gave the identical prompt, “Write an intro for a blog article on how to commencement a newsletter,” to the built-in Write a blog post accomplishment and to mine. The built-in accomplishment asked three questions concerning niche, audience, and spirit (I skipped them), projected an outline that opened alongside “Why newsletters are motionless valuable,” and wrote a three-paragraph intro that opens by marketing the idea of a newsletter.

Mine skipped the questions and opened alongside the reader: “You have ideas to share, but no apparent way to get them into readers’ inboxes without spending weeks wrestling alongside tools, templates, and signup forms.”

The accomplishment has limits, though. When I asked it to tighten a coarse paragraph alongside that “colleagues operating from extremely distinct places” comparison, it cleaned up the declaration about it but kept the difference as vague as it was. So I added a regulation telling it to create vague comparisons additional tangible and tried again. The paragraph came rear cleaner, but the difference was motionless fair as vague.

So the accomplishment handles the rules it can check, akin openings and dashes, but spotting a vague difference is motionless on me.

What I really do alongside AI now

None of this method penning without AI, but merely deciding which parts it gets and how to create the most of it. This is what that looks akin for me:

• Make the scheme before the prompt. I decide what a part argues and who it’s for before AI writes a sole paragraph, afterward form the outline alongside it until it’s mine. For this piece, I had the viewpoint and the meta tier in intellect before the archetypal outline, and I told the AI not to build one until I stated so.
• Let it discussion you. Before outlining this article, I had an AI ask me questions one at a time. Answering them brought up stories and examples I’d never have put in a prompt. The MacMagazine narrative that opens this part came out of a follow-up inquiry concerning my newsroom days.
• Feed it genuine material. Transcripts, notes, published pieces, your own examples. A generic topic gets a generic outline back. For this one, I shared AI drafts of two published pieces next to their final versions, and comparing them produced multiple of the examples you’ve peruse here.
• Do the final continue in your own space. Once a outline is close, I move it to Google Docs and reread all section, rewriting item that doesn’t audio akin me. This part went through exactly that, and you can difference it alongside the AI’s type in the discussion section.
• Open all origin it cites. When I looked up the base of the name nariz de cera, an AI-generated summary gave me a charming clarification but cited a origin that says item else. I additionally checked Graciliano’s citation against the first Portuguese before trusting our translation.
• Write the lines that matter yourself. The archetypal and final lines are anywhere readers decide what you audio like, so I compose quarry myself. For the ending, the AI suggested a few angles, and one of them stuck. The row itself is mine.

Which brings me rear to punctuation.

The one dash I kept

I motionless use em dashes, but lone whenever a declaration really needs one. Midway through my discussion alongside my AI assistant, I used one without equal noticing, in an answer concerning what matters most whenever I scheme an article.

The AI noticed. “You used a dash in that answer, by the way,” it wrote. “It earned its place.”

So if any of my old editors at MacMagazine are study this, yea — it’s motionless me.