We’ve been here before: what AI and self-publishing have in common

There’s a piece by Jane Friedman that I find myself returning to whenever the conversation around AI and writing starts to drift into extremes. She has already done the heavy lifting in terms of explaining what AI is, how it’s being used in publishing, and where the genuine concerns sit. If you haven’t read it, it’s worth your time: https://janefriedman.com/ai-and-publishing-faq-for-writers/

What I appreciate most about Jane’s approach is that it doesn’t fall into panic or blind enthusiasm. It holds space for both usefulness and caution, and that feels like the most honest place to sit. I absolutely agree with her.

AI can be incredibly useful for writers. Especially for those who never saw themselves as ‘writers’ in the first place, or for people who face barriers that make writing harder than it needs to be. Whether that’s confidence, structure, neurodivergence, time, or simply not knowing where to begin, AI can act as a bridge. And I don’t think we should be gatekeeping writing from anyone who wants to cross it.

Writing has always been a way of making sense of the world and being heard within it. If a tool helps someone do that, I’m not interested in shutting the door.

At the same time, for people who can already write, AI has a different kind of value. It becomes less about ‘doing the writing’ and more about supporting the thinking behind it. Organising ideas, sense-checking structure, exploring different angles, pulling threads together that might otherwise stay loose. Used well, it can sharpen what’s already there rather than replace it.

Where I start to lose patience is when the conversation shifts from judgement to surface-level suspicion.

I recently worked on a project where the author had become fixated on the idea that the use of an em-dash somehow signalled AI involvement. Their solution was to remove every single one during the editing process. The reasoning wasn’t about clarity or tone or flow. It was about optics. About not appearing to have used AI.

The irony is that the author had, in fact, involved AI. But that’s not the point.

Em-dashes didn’t arrive with AI. They’ve been part of strong, professional writing for years. They serve a purpose. They allow a sentence to breathe in a particular way; to pivot, to add nuance without breaking rhythm. Good writers have used them long before any language model existed.

AI, by its nature, is built on patterns. It learns from what already works. So it stands to reason that it will adopt the tools and techniques used by skilled writers. That includes punctuation choices.

Replacing em-dashes with something less suitable, purely to avoid suspicion, doesn’t improve the writing. It weakens it. And if we follow that logic through, we end up in a strange place where anything that becomes associated with AI gets stripped away. Today it’s em-dashes. Tomorrow it might be sentence length, tone, even clarity itself.

We don’t need to look too far ahead to see where that leads. Eventually, we’d run out of things we feel ‘allowed’ to use.

This is why I keep coming back to judgement.

We’ve seen this before. When self-publishing started to gain traction, there was a similar reaction. A sense that opening the gates would somehow dilute the quality of writing. That professional work would be overshadowed or devalued.

It didn’t happen.

What actually happened was that the market adjusted. Good self-published work found its audience. Poor work didn’t. Readers, as it turns out, are quite capable of deciding what’s worth their time.

I don’t see AI as fundamentally different in that respect.

If something is poorly written, disengaging, or lacks substance, it won’t be read, regardless of how it was produced. If something is thoughtful, compelling, or entertaining, people will find it. Whether AI played a role in shaping it becomes far less important than the experience of reading it. The content speaks.

There’s also an inconsistency in how we’re choosing to accept or reject these tools. Grammarly, for example, is widely used and rarely questioned. Yet mention tools like Claude or other generative AI platforms and the tone shifts. There’s a line being drawn, but it’s not always clear why it sits where it does.

Perhaps it’s familiarity. Perhaps it’s control. Or perhaps it’s just time.

If I were to make a prediction, and this is entirely my own, I don’t think this will be a particularly heated issue in ten years. Whether AI continues to evolve at pace or levels out into something more stable, it will likely become part of the landscape in the same way other writing tools have.

And when that happens, the focus returns to where it’s always been. The ideas. The voice. The ability to connect with a reader in a way that feels real.

That’s the part AI can’t fully replicate, not because it lacks the technical ability to form sentences, but because it isn’t living the experiences behind them.

So yes, use the tools. Explore them. Let them support you where they genuinely add value. (AI helped me organise my thoughts for this piece. No human feelings were harmed in the process.)