I’ve been listening to an artist called NOMARKMORE recently, so I was looking for their upcoming gigs. But I couldn’t find any – anywhere – which is pretty unusual for an up-and-coming band. So I started digging more into who they were. And the main thread I found online was wondering if they were AI (or not). Honestly, I’m still not 100% sure either way!
Here’s the relevant bit: the moment I couldn’t verify there was a real person behind the music, I started evaluating it differently. Same songs. Same production. Same everything I’d been enjoying five minutes earlier. But my brain had shifted from “this is great” to “wait, is this great?” Do I enjoy it as much if it’s AI-generated?
Why should the same output be evaluated differently based on its origin? It’s an interesting question for anyone who produces content or consumes it (which is just about everyone). Especially those who work in the field of AI.
Ultimately I still listen to them and enjoy ‘their’ music. The initial shock passed and I take the songs at face value now. But that moment of re-evaluation stuck with me, because it exposed something about how we’re all going to have to think about AI-generated work from here on.
The value discount
I think most people are going to apply a value discount to anything they suspect is AI. Not always consciously, and not always fatally, but it’s there. When you discover a piece of work was mostly machine-made, you look at it again with different eyes. You question whether the ideas are actually good or just plausibly-arranged. You wonder if the person putting their name to it did any real thinking.
I got sent a sales proposal recently from a well-regarded consultancy. Long document, all the right frameworks, plenty of pages. And I could tell within about 90 seconds it was 95% AI-written. Nothing specific I could point to, just the shape of the sentences, the way the argument flowed, the strange evenness of the whole thing. And once I’d clocked it, I discredited the entire proposal. Because if they hadn’t bothered to write it themselves, why would I bother to read it properly? What was the actual thinking they were bringing to the problem?
That’s the discount in action. It wasn’t that the proposal was bad on its own terms. It was that once I knew (or strongly suspected) the origin, the whole thing lost weight.
But then there’s the flip side. If I see someone’s LinkedIn post cover image is AI-generated, I don’t care. At all. We’re not all designers. It’s probably a better result than what they’d get from a webcam and a bit of natural light. The stakes are low, the intent is fine, and the image is largely ancillary to the content of their actual post.
So the discount isn’t universal. It’s context-dependent. And I think it comes down to three things.
When AI is fine, and when it isn’t
One There’s a real human origin story behind it.
Not that a human touched the final draft and removed some em dashes. The key here is that the thinking originated with a human. That there’s a point of view, a lived experience, a set of scars or opinions or observations that only that person could have produced. AI can help you express it, structure it, sharpen it, scale it. What it can’t do is generate it in the first place. Or rather, it can, but the output is the flat, generic, everyone-else-sounds-the-same stuff you’re already sick of scrolling past.
I write this way myself. My posts start as a spark, usually a thought from a call or something I’ve noticed. I record voice notes, save swipe files, look at the news and my social feed. This is all good inspiration.
Then I go through a collaborative process to sharpen the thinking, using Inklined to challenge and prod me to go deeper and avoid the lazy take. But the raw material, the point of view, the specific claim – that all comes from me. And I still add final edits anyway, because it’s my name on it.
That’s a legitimate use. It’s not different in kind from a founder who dictates to a ghostwriter, or an author who works with an editor. The origin is human. The execution has help.
Where it breaks down is when there’s no origin at all. When someone types “write me a LinkedIn post about ABM” into ChatGPT, hits publish, and calls it thought leadership. There’s no human at the source. There’s just a prompt and a generation. And you can feel that when you read it, even if you can’t put your finger on why.
Two The intent is to add value, not to fill space.
There’s a difference between using AI to make good work possible (I couldn’t have written this many articles without it) and using AI to churn out slop because volume is the metric. It’s the difference between leverage and pollution.
I had a conversation recently with someone whose platform explicitly positions itself as a ghostwriter rather than a one-shot content machine. The whole point being that by the time you finish the process, you’ve been part of the thinking. The output reflects you because you were in the loop. Compare that to the tools that will run an agent on your account, post for you, respond to comments as you, emulate your voice in DMs. That last one genuinely terrifies me. Not because the tech isn’t impressive, but because when the penny drops for the person on the other end that they’ve been talking to an AI wearing your face, the trust is gone. Not damaged. Gone.
Intent matters. And people can tell.
Three The quality is obviously high and time has genuinely been spent.
AI is still really bad at faking real quality. And when I say quality, I mean a specific thing: originality, a strong point of view, raw experience, unique learnings, fresh thinking. The stuff that comes from having actually done the thing.
You can spot the difference. A piece of writing that has a real argument, backed by specific examples the writer has actually lived, lands differently from a piece that has taken the same topic and produced the median take. Even if both are grammatically fine. Even if both hit the same word count. The signal is in the specifics…the small details, the counter-intuitive observations, the things you couldn’t have known unless you’d been in the room.
It’s like telling the difference in an interview between someone who’s tell a story about an experience they observed; versus one they actually lived.
If someone has genuinely spent time getting a piece as good as it can be, using AI to help, that’s fine. Great, actually. What isn’t fine is using AI to skip the time and effort, then pretending you did the work.
Why this matters for anyone building a brand
Everyone I talk to in GTM is wrestling with this in some form. How much AI is too much? Where’s the line? Am I going to get penalised for using it, or penalised for not using it fast enough?
Here’s my take. The market is getting flooded with AI-generated content on a scale we haven’t seen. Most of it is terrible. Some of it will be technically fine but forgettable. A small percentage will be genuinely good, and that small percentage will almost always be the stuff where a real human origin, real intent, and real quality-time have all been present.
If your content strategy is “use AI to produce more, faster, cheaper”, you’re competing in a race to zero. Everyone else has the same tools. The output converges. And your audience is developing better and better instincts for detecting it – the same instincts I applied to NOMARKMORE and to that consultancy proposal.
What can’t be commoditised is you. Your specific take, your accumulated experience, the pattern-recognition you’ve built up from years of doing the work. Layer AI on top of that and you get something that scales without losing what made it worth reading in the first place. Skip the human layer and try to get AI to generate it from scratch and you get the slop everyone’s learning to scroll past.
The people I know doing this well are using AI as an execution layer on top of a real base of knowledge and a real point of view. Recorded conversations, notes from client calls, actual opinions formed over years of practice. All of that gets structured and expanded and shaped with AI. But the raw material is theirs. It’s specific. It couldn’t have come from anyone else.
The value discount, revisited
Coming back to NOMARKMORE. I still don’t know if they’re AI. And after the initial shock, I’ve mostly stopped caring, because the music is good enough that it stands on its own. That’s the pass condition. When the final output is genuinely great, and there’s at least a sense of authenticity underneath it, most people will forgive AI in the workflow.
But that’s a high bar. Higher than most people are meeting. And the discount is going to get harsher, not gentler, as audiences get more sensitive to the tells.
So the question I’d be asking, if I were building anything under my own name right now, isn’t “should I use AI?” It’s “would this pass the sniff test if someone found out exactly how it was made?”
If yes, carry on. If no, the AI isn’t the problem. The absence of you is.