An admission. Lynne Walley asked how much of my recent writing is actually me, and how much is ChatGPT

An admission. Lynne Walley asked how much of my recent writing is actually me, and how much is ChatGPT. Join the conversation on LinkedIn. https://www.linkedin.com/posts/antlerboy_i-read-scott-odells-island-of-the-blue-activity-7460599789932548096-BFTs?utm_source=share&utm_medium=member_desktop&rcm=ACoAAACuq-oBecVFDW6PCf3lkoG-peMeuLBeoho

An admission. Lynne Walley asked how much of my recent writing is actually me, and how much is ChatGPT. Honestly: more than I would’ve admitted a year ago. Maybe 75% me, 25% AI on average. Though it varies hugely.

A recent post started as a WhatsApp conversation with my mum about a book she’d reread and apologised for giving me when I was five because she’d suddenly realised it was absurdly complicated for a child.

I used Google and ChatGPT to remind myself of names and plot points.

Turned the chat into notes, identified what I wanted to say.

Asked ChatGPT to reshape it into a LinkedIn-post-shaped thing.

Then rewrote large chunks because it got repetitive, sentimental, and sounded alarmingly like ChatGPT.

Which is increasingly what my workflow looks like.


I don’t really experience this as ‘AI writing for me’. It’s more like another component in a cognitive ecology that already included:

– Roam Research database

– old blog posts

– slides

– search engines

– books!

– my own archives (including my ‘80% done’ book draft)

– the LinkedIn algorithm.

And now language models. A kind of cybernetic relationship.


What unsettles me is that some of the things I thought I was pretty good at – synthesis, pattern-spotting, connecting ideas – are exactly the things these systems are surprisingly good at. And (sub)editing – but not wisdom, not judgement, not actually writing and redrafting anything longer.


At the same time, whenever I really care about meaning, I still end up doing the final integration myself.


What I like is when AI helps people become more themselves, do synthesis that wouldn’t otherwise have happened. What deadens me is the flattened ‘ChatGPT voice’ – which, horrifyingly, one can slowly begin to acquire oneself.


And underneath all this sits the bigger thing: the accelerating river of content, which I am obviously contributing to too.


Anyway. Mild mea culpa – but I’m happy to have this out there!

And NB this post was based on my response here:

https://lnkd.in/eVaMtz6p …with AI assitance!