Every conversation about AI and writing seems to collapse into two camps.
One says AI is corrosive to authentic voice.
Use it and you’re diluting yourself.
The other says AI is leverage.
Prompt it well, tell it to “write in my voice,” publish more, scale faster.
The gurus selling that second position are, more often than not, also selling the course or the software that makes it look effortless.
Complexity doesn’t sell software. A three-step framework does.
I don’t belong to either camp.
Because both are arguing about the wrong thing.
The question isn’t whether you use AI.
It’s what you’re willing to outsource to it.
Judgement versus periphery
I wrote a note yesterday.
I don’t outsource my truth. I outsource my blind spots.
A reader, Jenni, reflected it back with another phrase: peripheral vision.
That stopped me.
Because it named the distinction I’d been reaching for.
Judgement versus periphery.
Judgement is the part of you that decides what’s true, what matters, what you actually think.
Periphery is simply what you can’t see from where you’re standing.
Every writer has one.
Every leader has one.
Blind spots aren’t a flaw to be ashamed of.
They’re the cost of having a specific point of view at all.
Outsourcing judgement to AI means letting it decide what you think, then wearing the output as if it were yours.
That’s not collaboration.
It’s borrowed judgement wearing your name.
Outsource the periphery, not the judgement
Outsourcing peripheral vision means using AI to see what you structurally cannot.
To stress-test an argument.
To surface the counter-case you’re too invested to notice.
To catch the place your reasoning quietly assumed something you never examined.
That’s collaboration.
Your judgement stays intact.
Arguably sharper, because it’s been tested against something other than your own reflection.
What “write in my voice” leaves out
Real collaboration is expensive in a currency most people aren’t willing to spend.
Time.
Iteration.
The discomfort of being told your first draft’s thesis doesn’t hold.
It’s closer to training a junior collaborator than issuing an instruction.
You don’t get peripheral vision from “sound like me.”
You get it by bringing your actual thinking into the conversation — your positions, your contradictions, your unresolved questions — and asking the model to push back against them.
It’s a practice.
Strip the mystique off it and it’s unglamorous.
You don’t open with a prompt.
You open with a position — something you already believe, worked out the hard way, before the model was ever in the room.
Then you ask it to do the one thing a mirror can’t.
Disagree with you on the evidence.
Where does this contradict something I said last month?
What’s the counter-case a smart critic would raise first?
Where did I skip a step because it felt obvious to me and isn’t obvious at all?
Most of what comes back gets discarded.
You’re not looking for material.
You’re looking for the gap between what you meant and what you actually wrote down.
The output that survives reads as more you, not less.
Because everything that wasn’t rigorously yours has already been cut.
Used this way, AI isn’t primarily adding.
It’s helping you remove what doesn’t hold.
The cliché.
The borrowed argument.
The contradiction you hadn’t noticed.
The sentence that sounds intelligent but isn’t actually yours.
That’s a very different use of the technology from asking it to manufacture a voice.
The technology isn’t the variable
The binary camp of pro-AI, anti-AI treats the technology as the variable.
It isn’t.
The variable is whether you’re still the one making the call.
A leader who knows their own judgement can hand a model their blind spots without losing themselves in the exchange.
Because they know what’s theirs before the model ever gets involved.
A leader who hasn’t done that work has nothing stable to check the output against.
AI doesn’t create that instability.
It just makes it visible faster, and at higher volume, than a human editor ever could.
So the real question isn’t whether you should use AI in your writing.
It’s whether you know your own judgement well enough to recognise when you’ve stopped using it.
AI can challenge your thinking.
It can expose contradictions.
It can widen your peripheral vision.
But it cannot tell you what is yours.
That judgement has to remain with you.
Perhaps that’s the more interesting divide we’re heading towards.
Not people who use AI and people who don’t.
People who use it to strengthen their judgement and people who use it to replace it.
If you’ve read this far unsure which side of that line you’re actually on, that’s exactly where Signal Fire begins.
Not by telling you what to think.
By helping you see, clearly, what’s your judgement and what’s borrowed.
You already know more than you think you do.
The difficult part is recognising what has distorted your judgement.
And what becomes possible once you can see it clearly.
Signal Fire is a private one-to-one conversation for quietly capable people carrying responsibility, decisions or identities that no longer feel like their own.
I’ve found myself wrestling with the difference between judgment and what sits on the edges of our awareness. That distinction matters more than we often realize.
When it comes to growing as leaders, I’ve noticed that the blind spots that trip us up are usually the ones we’re convinced we’ve already handled. Lately, I’ve been experimenting with AI—not to do the work for me, but to challenge my thinking. Treating AI like a sparring partner helps me see where my assumptions need work. It doesn’t take away my responsibility as a leader. If anything, it pushes me to get clearer about what really matters.
I’m grateful for this fresh way of looking at collaboration. It reminds me that real partnership is about sharpening each other, not just agreeing.