ai;dr and the new era of writing

How LLMs have made our writing … different. And what to do about it

Hi there! You’re reading the Bonfire newsletter from Kevan Lee & Shannon Deep. Each week, we highlight learnings from our experience as in-house marketers turned agency owners who think a lot about creativity, our relationship to work, and how all of that impacts our identities.

Wishing you a great week!

---

AI’s impact on the way we create

When I sat down to write this newsletter, my thoughts fell onto a somewhat new and dangerous track:

• Ooh, I have a newsletter idea!

• Hmm, the idea is only a headline.

• Ah, I think this is where I want to take it.

• Writing is hard.

• What if I had ChatGPT do it?

Rest assured, I did not capitulate.

I wrote the whole thing myself, as I do every time.

But the temptation was there.

This was not always my approach to writing, wanting someone else to do it for me. Though writing may be hard, writing is also really fun. It is one of my favorite things! When I was working in-house as a content marketer, I used to write four 1,500-word blog posts per week about how to build your business’s social media strategy for Facebook Pages, Twitter, and Google Plus (imagine telling my grandkids that!), and I could not have imagined a more luxurious role. The job was dreamy.

Of course, that job is a relic now. No human person is writing four 1,500-word blog posts per week, from scratch, all by themselves. Many people are not even writing their own emails anymore.

Today, writing is a joint exercise between the AIs and us.

Are we, as writers, better off because of it?

Are you, as readers, destined to consume slop?

How has it changed our habits as writers and readers and creators?

The topic has been on my mind lately as I’ve noticed my own media habits evolving quickly, my tastes getting more refined, and my workday habits—writing and reading included—teetering on the edge of AI transformation.

Where do you draw the line with AI’s involvement in content? I’d love to hear your thoughts. Here are some of mine.

We are savvier readers because of AI

Marketing has a long history of people developing an aversion to marketing. Consumers developed “banner blindness” by no longer noticing or clicking on digital ad banners. Communities can sniff out an insincere promotional message in seconds.

We’ve reached the same high alert to AI nonsense.

There’s even a term for it: ai;dr, which stands for AI-written, so I didn’t read it, a version of the popular tl;dr abbreviation that stands for too long, didn’t read.

Someone shares a quote with you that’s riddled with em-dashes?

That’s ai;dr.

Some email or blog post goes viral with the weirdest, most unnatural sentence style imaginable?

Instant ai;dr.

The origins of ai;dr come from social media, where an AI safety researcher’s resignation letter went viral on Threads for being suspiciously cringe, as if written by a bot and not the researcher. Here’s what Carmen Vicente at Scroll Sick found:

In the comments a developer named David Minnigerode replied, “Sorry, that is definitely tl;dr. But also kinda ai;dr. Some of those sentences…yeesh.” In layman’s terms, this is AI slop and I don’t wanna read it.

#2: Unexpected Tears, and The Good Kind of Quitting

Carmen goes on to say:

I’d rather read a poorly written, poorly structured essay from a real person, than polished genAI.

What’s implied here is that much of AI-created content is noticeably bland—the same tropes and devices used over and over again, like em-dashes and imperative sentences—or distractingly bizarre—off-topic asides or prompts taken to an extreme. Of course, some AI content creators are really great at building effective prompts and GPTs of their own, but the majority of people aren’t operating as sophisticatedly and it shows in the large volume of inauthentic content, particularly published in the business world either by brands or by influencers.

I’ve noticed this for myself when I’m reading longer-form work like:

• Blog posts

• LinkedIn content

• Newsletters

• Strategic docs and memos

• Emails

It more easily escapes my notice on short-form content like:

• Lists

• Headlines

• Subject lines

As AI has proliferated, our AI radar has improved where we’re able to sense when something feels “off” with content. How long will this last before the AI improves or the prompting gets better? Hard to say. What does this mean for content creators? There’s a big chance to stand out by being you … and sounding like you.

We need to be savvier writers, working with AI

AI is very good at certain layers of writing:

• Structuring

• Summarizing

• Pattern-matching

• Filling in gaps

• Generating options

But those were never the hard parts.

The hard parts are:

• Having a point of view

• Noticing something others haven’t articulated yet

• Deciding what matters

• Choosing what not to say

Of course, the hard parts are the exact parts people are most tempted to outsource, which leads to a weird inversion: We keep the easy work (editing, polishing, tweaking tone), and hand off the hard work (thinking).

This is why we end up with bland, bizarre, and unoriginal AI content that human readers can smell a mile away.

AI should be helping us be more productive (in theory), so the question for writers and creators and content marketers becomes how and where to integrate AI into the writing process.

To reflect back on my days of writing four 1,500-word blog posts per week, my non-AI process was a three-day, staggered workflow where in a single day I’d write a first draft of a new story, do a final edit of a draft written the day before, and publish the draft from two days before. Then the following day, I’d move everything up one stage: write a new draft, edit a second one, and publish and promote a third one.

If I had the luxury of generative AI back then, I could have applied it to the easier work of

• Revising my outline for structure

• Summarizing my research

• Pattern-matching with other ideas I’d explored before

• Filling in gaps in my reasoning

• Generating options for headlines and keywords

Rather than prompting AI with:

I need 25 ideas for blog posts about Facebook marketing. Come up with this list, and write me a first draft of the best idea.

I could have used:

Read the following draft and identify any gaps in reasoning or opportunities to support my arguments with more research

We’re still finding our ideal AI workflow at Bonfire. (Any tips you all have would be much appreciated!) But it’s not too far off from what I’ve described above: We are starting our work with human perspective, using AI to further refine, and then ending with a final human edit before anything goes live. It’s a human-AI-human sandwich.

We start with an idea, a direction, a concept, or a draft, conceived by us humans.

Then we hand off to AI to give us an outline, riff on more options, or propose a basic draft.

Then we human-edit the AI content so that it is accurate, authentic, and true to our original vision.

How to move forward

With the promise of AI productivity more tempting than ever, it’s no wonder that bad AI content is on the rise. More people are using AI for more things, and the quality has not caught up. And people notice the divide between AI-created and human-created content.

As a writer and creator both for pleasure and for work, I’ve felt tugged in all sorts of directions when it comes to how I use AI, how I write, and how I create. I try to use AI in a way that preserves the parts of writing stories and making content that actually feel meaningful.

Here’s my latest shortlist of rules and reminders, which I’m sure will continue to evolve:

• Use AI after the thinking is done, not before.

• AI is great at helping shape something. It is not great at deciding what the something should be.

• I ask it to give me options, poke holes, suggest structures. I won’t ask it to “write the thing.”

• For speed, not substance

• Tightening, summarizing, reformatting, repurposing: the mechanical parts of research and writing that take me longer than they should.

• Sometimes I just want to see my idea reflected back in a different way so I can react to it and agree or disagree or refine.

What rules and reminders have you put in place?

It’d be great to hear from you in the comments or replies!

---

But wait! There’s more…

Wanna hang out in person?

Our next retreat this month (happening right now, actually) is SOLD OUT! However, you can sign up for the mailing list to find out when the next one is happening.

Wanna work with us?

If you need help with brand strategy and storytelling, fractional brand and marketing leadership, and bringing your brand strategy to life in impactful ways, send us an email at hello@aroundthebonfire.com to get in touch.

As always, you can find us on LinkedIn, Instagram, and Threads.